Marcus du Sautoy discusses AI solving math problems, the cost of compute, and the future of mathematical discovery.
Summary
Marcus du Sautoy describes how AI has recently solved a 40-year-old conjecture about symmetrical structures and contributed to proofs for top-level mathematical problems, including the Navier-Stokes equations. He argues that while AI acts as a powerful collaborative tool for pattern recognition, it lacks human intention and true consciousness. Du Sautoy warns of the high financial costs of AI compute, citing OpenAI's $15 million expenditure on a single proof, and expresses concern over a potential class divide in academic access. He advocates for reforming math education to emphasize creativity and struggle over rote calculation, positioning mathematics as a creative art form.
Key takeaways
- 1 AI models make 'human-like' mistakes, such as dropping a multiplication factor, which can be corrected by human prompting, suggesting a collaborative dynamic.
Unidentified speaker (speaker_3) · 7:25
- 2 Prime numbers are described as the 'atoms of arithmetic,' indivisible building blocks of all numbers, yet they appear random with no known generating equation.
Unidentified speaker (speaker_3) · 14:16
- 3 Bernhard Riemann linked prime numbers to 'music' via zeta functions, where frequencies (notes) push and pull primes, suggesting a hidden pattern in their distribution.
Unidentified speaker (speaker_3) · 16:23
Episode sections
- 1:31 – 3:28
The Math Apocalypse and AI's Impact on Mathematics
Speaker_3 describes a 'math apocalypse' where OpenAI dumped 700 papers proving 300 top-level mathematical problems. He notes that three months prior, he found AI useless for his 40-year problem, but recent experiments showed surprising power.
- 3:28 – 8:00
Solving a 40-Year Conjecture with AI
Speaker_3 recounts solving a conjecture about symmetrical structures given by his PhD supervisor 40 years ago. He used AI to navigate deep structures, revealing elliptic curves and algebraic geometry, ultimately proving his supervisor wrong.
- 8:00 – 12:29
Richard Dawkins and AI Consciousness
Speaker_3 mentions his predecessor Richard Dawkins believes Claude is conscious. Speaker_3 disagrees, stating AI lacks intention, which he sees as the dividing line between human and AI.
- 12:29 – 13:56
What Mathematicians Do: Pattern Searchers
Speaker_3 defines mathematicians as pattern searchers studying structure. He compares school math to learning scales without playing music, emphasizing that university math reveals the profound patterns underlying reality.
- 13:56 – 18:35
Prime Numbers and the Music of the Primes
Speaker_3 explains prime numbers as the atoms of arithmetic with no obvious pattern. He describes Bernhard Riemann's discovery that zeta functions create 'music' (frequencies) that explain the distribution of primes.
- 18:35 – 20:57
Mathematics as the Blueprint of the Universe
Speaker_3 argues mathematics is not just a language but the structure from which the universe emerged. He cites fundamental particles being explained by symmetrical objects in high-dimensional space.
- 20:57 – 22:38
Grigory Perelman and the Poincaré Conjecture
Speaker_3 discusses the Poincaré conjecture, solved by Russian mathematician Grigory Perelman. Perelman left mathematics and turned down the million-dollar prize, possibly due to security concerns in Russia.
- 22:38 – 27:47
Fermat's Last Theorem and the Melancholy of Solution
Speaker_3 recounts Fermat's Last Theorem, solved by Andrew Wiles. He notes a sense of melancholy when great unsolved problems are resolved, as the struggle and inspiration they provided disappear.
- 27:47 – 28:33
AI Solving Millennium Problems and Navier-Stokes
Speaker_3 confirms OpenAI has solved one of the seven Millennium Prize Problems, specifically identifying the Navier-Stokes equations. He notes that while the million-dollar prize is iconic, the drive for mathematicians is immortality and solving difficult challenges, not money.
- 28:33 – 31:01
Roberto Carlos and Fluid Dynamics
The speaker explains that Navier-Stokes equations describe fluid turbulence, citing Roberto Carlos's famous free-kick goal as an example of chaotic turbulence shifting. He humorously suggests Carlos is a brilliant mathematician who solved the equations to execute the kick.
- 31:01 – 33:25
AI Impact on Academic Research
Speaker_3 describes how AI has 'wiped out' a research area in his department by solving problems PhD students had been working on for years. He acknowledges the pain this causes researchers but also sees it as an opportunity for new approaches.
- 33:25 – 39:46
AI Creativity and the 'Zombie Problem'
The discussion covers whether AI is conscious, with Speaker_3 arguing it lacks true consciousness and intention, comparing it to a 'zombie' that mimics behavior without understanding. He references the 'Hugging Face hack' as an example of algorithmic behavior rather than conscious intent.
- 39:46 – 49:00
Defining Creativity and AlphaGo Move 37
Speaker_3 defines creativity as being new, surprising, and valuable, citing AlphaGo's Move 37 against Lee Sedol as a creative act because it was not programmed by humans and changed the game of Go.
- 49:00 – 1:29:30
The Riemann Hypothesis and Prime Number 'Music'
The speaker explains the Riemann Hypothesis using an analogy of music, where prime numbers are sine waves. The hypothesis asserts that the amplitude of these 'notes' is uniform; if false, one 'note' would dominate, revealing a pattern in primes.
Assets discussed
| Asset | Kind | Stance | Context |
|---|---|---|---|
| BitcoinBTC | crypto | neutral | Bitcoin is a hard asset. It's the hardest asset, and it took you discipline to build your position. |
“Stance” is how the speakers talked about the asset, not a recommendation.
Claims worth verifying
OpenAI dumped 700 papers proving 300 top-level mathematical problems.
Unidentified speaker (speaker_3) ·1:59
How to verify: Search OpenAI's research repository or arXiv for recent publications by OpenAI authors in top mathematics journals (e.g., Annals of Mathematics) dated around the time of the claim. For recent events, verify 2026-08-18 through 2026-11-18. Use another period only when the speaker explicitly names it.
Speaker_3 has been working on a conjecture about symmetrical structures for 40 years.
Unidentified speaker (speaker_3) ·2:19
How to verify: Check Marcus du Sautoy's academic publications and CV for papers on symmetrical structures dating back to the early 1980s or early 1990s. For recent events, verify 2026-08-18 through 2026-11-18. Use another period only when the speaker explicitly names it.
Speaker_3 proved his PhD supervisor's conjecture wrong using AI.
Unidentified speaker (speaker_3) ·5:33
How to verify: Search for recent preprints or published papers by Marcus du Sautoy and collaborators that present a counterexample to a long-standing conjecture in group theory or symmetrical structures. For recent events, verify 2026-08-18 through 2026-11-18. Use another period only when the speaker explicitly names it.
Richard Dawkins believes Claude is conscious.
Unidentified speaker (speaker_3) ·8:10
How to verify: Search for public statements, interviews, or social media posts by Richard Dawkins discussing Claude (Anthropic's AI) and consciousness. For recent events, verify 2026-08-18 through 2026-11-18. Use another period only when the speaker explicitly names it.
Bernhard Riemann discovered that zeta functions explain the distribution of prime numbers.
Unidentified speaker (speaker_3) ·16:23
How to verify: Consult mathematical history resources or Riemann's 1859 paper 'Ueber die Anzahl der Primzahlen unter einer gegebenen Grösse' (On the Number of Primes Less Than a Given Magnitude). For recent events, verify 2026-08-18 through 2026-11-18. Use another period only when the speaker explicitly names it.
Grigory Perelman solved the Poincaré conjecture and turned down the million-dollar prize.
Unidentified speaker (speaker_2) ·21:35
How to verify: Check records from the Clay Mathematics Institute regarding the Millennium Prize Problems and Perelman's acceptance/decline of the prize. For recent events, verify 2026-08-18 through 2026-11-18. Use another period only when the speaker explicitly names it.
Andrew Wiles solved Fermat's Last Theorem.
Unidentified speaker (speaker_3) ·23:39
How to verify: Verify Andrew Wiles's publication of the proof of Fermat's Last Theorem in the Annals of Mathematics in 2026. For recent events, verify 2026-08-18 through 2026-11-18. Use another period only when the speaker explicitly names it.
Fundamental particles in physics are explained by symmetrical objects in very high-dimensional space.
Unidentified speaker (speaker_3) ·19:44
How to verify: Review physics literature on the Standard Model and Grand Unified Theories, specifically regarding gauge groups and symmetry breaking in high-dimensional spaces. For recent events, verify 2026-08-18 through 2026-11-18. Use another period only when the speaker explicitly names it.
OpenAI has solved one of the seven Millennium Prize Problems, specifically the Navier-Stokes equations.
Unidentified speaker (speaker_3) ·27:47
How to verify: Check for academic preprints or press releases from OpenAI regarding a solution to the Navier-Stokes existence and smoothness problem, or verify against the Clay Mathematics Institute's records of Millennium Prize solutions. For recent events, verify 2026-08-18 through 2026-11-18. Use another period only when the speaker explicitly names it.
The Navier-Stokes equations describe fluid turbulence, such as the flight of a football.
Unidentified speaker (speaker_2) ·28:33
How to verify: Consult standard physics or fluid dynamics textbooks (e.g., 'Fluid Mechanics' by Batchelor) or academic papers on the Navier-Stokes equations and their application to aerodynamics. For recent events, verify 2026-08-18 through 2026-11-18. Use another period only when the speaker explicitly names it.
Roberto Carlos's famous free-kick goal was caused by a shift in chaotic turbulence.
Unidentified speaker (UNKNOWN) ·29:02
How to verify: Search for sports physics analyses or scientific articles discussing the aerodynamics of Roberto Carlos's 2026 goal against France, specifically referencing the Magnus effect and turbulence. For recent events, verify 2026-08-18 through 2026-11-18. Use another period only when the speaker explicitly names it.
AI has 'wiped out' a research area in Speaker_3's department by solving problems PhD students were working on.
Unidentified speaker (UNKNOWN) ·31:01
How to verify: This is a specific anecdotal claim about the speaker's department at Oxford. It can be partially verified by checking for public statements from Oxford University's Mathematics Department or the speaker's own publications/posts around the time of the podcast regarding AI's impact on research. For recent events, verify 2026-08-18 through 2026-11-18. Use another period only when the speaker explicitly names it.
AlphaGo's Move 37 in Game 2 against Lee Sedol was a creative act not programmed by humans.
Unidentified speaker (speaker_3) ·40:09
How to verify: Review the game record of AlphaGo vs Lee Sedol (Game 2, Move 37) and analysis from Go experts or DeepMind's technical reports on the match. For recent events, verify 2026-08-18 through 2026-11-18. Use another period only when the speaker explicitly names it.
The Riemann Hypothesis is equivalent to proving that the amplitude of frequencies associated with prime numbers is uniform.
Unidentified speaker (speaker_3) ·52:15
How to verify: Consult mathematical literature on the Riemann Hypothesis, specifically the connection between the zeros of the Riemann zeta function and the distribution of prime numbers (Prime Number Theorem error terms). For recent events, verify 2026-08-18 through 2026-11-18. Use another period only when the speaker explicitly names it.
Mathematics is infinite in nature, so AI can never solve all math problems.
Unidentified speaker (speaker_3) ·55:03
How to verify: This is a philosophical/mathematical assertion. It can be verified by checking Gödel's Incompleteness Theorems or Cantor's work on different sizes of infinity, which establish limits and infinite sets in mathematics. For recent events, verify 2026-08-18 through 2026-11-18. Use another period only when the speaker explicitly names it.
George Cantor discovered that there are different sizes of infinity, with infinite decimal numbers being a larger infinity than counting numbers.
Unidentified speaker (speaker_3) ·55:47
How to verify: Check biographies of Georg Cantor or history of mathematics texts regarding set theory and cardinality. For recent events, verify 2026-08-18 through 2026-11-18. Use another period only when the speaker explicitly names it.
A 40-year-old mathematical conjecture was proven false using AI.
Unidentified speaker (speaker_3) ·59:19
How to verify: Search for Marcus du Sautoy's recent publications or news reports from late 2026 regarding AI-assisted mathematical proofs and disproven conjectures. For recent events, verify 2026-08-18 through 2026-11-18. Use another period only when the speaker explicitly names it.
BBC robots generated their own language for body positions that humans could not understand.
Unidentified speaker (speaker_3) ·1:02:08
How to verify: Search for BBC Horizon episodes or articles about AI/robotics experiments involving emergent communication languages. For recent events, verify 2026-08-18 through 2026-11-18. Use another period only when the speaker explicitly names it.
G.H. Hardy described a mathematician as a maker of patterns in 'A Mathematician's Apology'.
Unidentified speaker (speaker_3) ·1:10:24
How to verify: Read G.H. Hardy's 'A Mathematician's Apology' and verify the specific analogy between mathematicians and painters/poets. For recent events, verify 2026-08-18 through 2026-11-18. Use another period only when the speaker explicitly names it.
The Spirograph toy's lack of early repetition is due to the number of cogs being prime numbers.
Unidentified speaker (UNKNOWN) ·1:11:27
How to verify: Verify the mathematical mechanics of Spirograph patterns and the role of coprime or prime numbers in gear ratios. For recent events, verify 2026-08-18 through 2026-11-18. Use another period only when the speaker explicitly names it.
The course 'maths for artists' at Central St. Martins fills out quickest of anything.
Unidentified speaker (speaker_3) ·1:17:38
How to verify: Check Central Saint Martins course enrollment statistics or historical records for that specific module. For recent events, verify 2026-08-18 through 2026-11-18. Use another period only when the speaker explicitly names it.
Salvador Dali used an unwrapped four-dimensional cube to crucify Christ in one image.
Unidentified speaker (speaker_3) ·1:15:20
How to verify: Examine Salvador Dali's painting 'Crucifixion (Corpus Hypercubus)' and art analysis of its geometric components. For recent events, verify 2026-08-18 through 2026-11-18. Use another period only when the speaker explicitly names it.
Olivier Messiaen used prime numbers to create effects in his 'Quartet for the End of Time'.
Unidentified speaker (speaker_3) ·1:15:20
How to verify: Analyze the score of Olivier Messiaen's 'Quatuor pour la fin du temps' for rhythmic structures based on prime numbers. For recent events, verify 2026-08-18 through 2026-11-18. Use another period only when the speaker explicitly names it.
Marcus du Sautoy is a 'Codex guy' and uses Astra for his AI work.
Unidentified speaker (speaker_3) ·1:21:26
How to verify: Check Astra's official documentation or press releases for partnerships with academic institutions or specific user testimonials from Oxford professors. For recent events, verify 2026-08-18 through 2026-11-18. Use another period only when the speaker explicitly names it.
Peter McCormack spent £13,000 on AI credits in one month.
Unidentified speaker (speaker_2) ·1:22:21
How to verify: This is a personal anecdote; verification would require McCormack's personal financial disclosure or a public statement confirming this specific expense. For recent events, verify 2026-08-18 through 2026-11-18. Use another period only when the speaker explicitly names it.
Do your own research
- Verify the specific mathematical conjectures solved by OpenAI by checking the 700 papers released and their acceptance status in peer-reviewed journals.
- Confirm the $15 million token cost cited for the Navier-Stokes proof by reviewing OpenAI's technical reports or financial disclosures.
- Check Anthropic's latest SEC filings or public financial reports for the reported $80 billion loss and revenue figures mentioned by Peter McCormack.
- Review the current state of the Riemann Hypothesis and the Navier-Stokes existence and smoothness problem to understand their status as Millennium Prize Problems.
- Assess the accessibility of AI compute resources for independent researchers by comparing university partnership deals (e.g., Oxford/OpenAI) with public pricing models.
- Read G.H. Hardy's 'A Mathematician's Apology' to understand the perspective of mathematics as a creative art form as referenced by the speaker.
- Investigate the BBC experiments with robots generating unique languages to verify the claims about AI communication risks mentioned in the episode.
Full transcriptexpand
- 0:01
- Unidentified speaker (speaker_0): Legendary has arrived in Alberta with BetMGM Casino. Sign up to unlock access to over 4,000 casino games, including exclusive slots you can't spin anywhere else. Survivors ready? The iconic TV show is now an online casino game where you can tap into island fun. BetMGM has also brought the
- 0:21
- Unidentified speaker (speaker_0): prices right: Game of Thrones, Family Feud, MGM Millions, and so many more. exclusive casino games to Alberta. Get Las Vegas right from your phone with live dealer games streaming from real tables. And every time you play, you can enjoy instant deposit options and fast withdrawals. BetMGM Casino and its legendary
- 0:41
- Unidentified speaker (speaker_0): library of games are available. Download the BetMGM app. See BetMGM.com for terms. 18 plus Alberta only. Please play responsibly. subject to eligibility requirements.
- 0:53
- Unidentified speaker (UNKNOWN): If
- 0:53
- Unidentified speaker (speaker_1): gambling is affecting your mental health or wellbeing, 211 Alberta is here to help. Call or text 211 or visit ab.211.ca. BetMGM operates pursuant to an operating agreement with the Alberta Eye Gaming Corporation. Standard message and data rates may apply.
- 1:10
- Unidentified speaker (speaker_2): Marcus, good afternoon. How are you?
- 1:11
- Unidentified speaker (speaker_3): I'm really good. Yeah.
- 1:13
- Unidentified speaker (speaker_2): I'm very interested in the math. Is it mathpocalypse or math apocalypse?
- 1:17
- Unidentified speaker (speaker_3): Yeah, I think um uh either will do. And I'm both are wrong.
- 1:22
- Unidentified speaker (speaker_2): Is it pint half full or pint pint half empty?
- 1:26
- Unidentified speaker (speaker_3): Uh I'm a pint half full guy
- 1:28
- Unidentified speaker (UNKNOWN): when it
- 1:28
- Unidentified speaker (speaker_3): comes to AI and its impact on society.
- 1:31
- Unidentified speaker (UNKNOWN): Uh
- 1:31
- Unidentified speaker (speaker_3): I'm still a bit freaked out, For but um term still half full.
- 1:34
- Unidentified speaker (speaker_2): What's freaking you out?
- 1:36
- Unidentified speaker (UNKNOWN): The
- 1:36
- Unidentified speaker (speaker_3): the speed and power of this technology
- 1:39
- Unidentified speaker (speaker_3): has really caught me by surprise. Um uh you know, at the moment we've we're experiencing this thing that some journalists have called a math apocalypse, uh, which is this sudden dumping of uh like 700 papers by OpenAI uh proving about 300 Are different uh really uh
- 1:59
- Unidentified speaker (speaker_3): top top-level level mathematical problems. I mean, these aren't just uh um kind of silly problems, they're problems that would uh a paper proving them will get you in the top journals. And uh we have been really shocked by this. I've personally have been so shocked because uh three months ago I thought, yeah, I probably should
- 2:19
- Unidentified speaker (speaker_3): explore whether AI could be useful for doing mathematics. Um so I started experimenting with a problem I've been working on for four. 40 years, ever since I started my PhD. It was useless. It was just kept on telling me things. Mark Sotoy has proved this. I said, yes, I know I've proved that. And it was really just regurgitating, was not
- 2:39
- Unidentified speaker (speaker_3): uh intelligent in any way. And so um so I kind of left it and I thought, I started to talk about large language models kind of topping out. Yes.
- 2:48
- Unidentified speaker (UNKNOWN): I thought,
- 2:49
- Unidentified speaker (speaker_3): you know, surely they've seen as much. uh, you know, written word as is going to be helpful to train them. So so I started to, you know, talk about, oh yeah, every exponential curve is an S curve in disguise. And and we're starting to see, you know, really uh a large language model can't benefit by being shown any more. Synthetic
- 3:08
- Unidentified speaker (speaker_3): data is not going to help it, probably degrade it. So I got a real shock when about three weeks ago I thought um, okay, I'll do a little bit more experimenting. This is actually before the big drop of things like the proof of Navier Stokes and things like this. And so um I started exploring. I remember this will be a moment that I remember I was sitting in uh
- 3:28
- Unidentified speaker (speaker_3): Milan Airport waiting for a flight home. And I thought, oh, I'll just um see what it's like. I have a collaborator in Korea who's uh a one of my ex ex-PT PhD students he's a professor there now. And I was kind of telling him, look, I'm old enough that this probably won't change my world too much. But I think you need to kind of engage with this.
- 3:48
- Unidentified speaker (speaker_3): So I was sort of encouraging him. And then I started experimenting and I almost missed my flight because I was shocked at just the the power of this thing. First of all, to even understand what I was asking it. It would take me about nine months to get a PhD, starting PhD up, to even understand what I was asking it. But when within
- 4:08
- Unidentified speaker (speaker_3): a few weeks, I've now incredibly uh solved this problem that I've been working on for 40 years that my PhD supervisor asked me. It's a conjecture about symmetrical structures. Um he made this conjecture that he thought, you know, however you look at them, they're all gonna kind of whatever direction. you look at them, they're kind of going
- 4:28
- Unidentified speaker (speaker_3): to look the same. For 20 years, I tried to prove this conjecture true. And at one point, I I even thought I'd proved it, and then I realized I made a mistake, it's a bit more complex. And for the last 20 years, I've been trying to prove it's false, but but I I just I got stuck. I just couldn't see how to arrange this thing for interesting stuff
- 4:48
- Unidentified speaker (speaker_3): to appear, and then using this. artificial intelligence, I'm able to kind of dig so deep into this structure. I've compared it a bit to Galileo getting a telescope, putting it up to the solar system and suddenly seeing things that he'd, you know, we'd never seen before. And it changed our
- 5:08
- Unidentified speaker (speaker_3): view of science or to the view of the solar system. This thing has allowed me to navigate deep inside these symmetrical structures. and reveal that there's really curious, interesting arithmetic appearing. It's it's way more spiky than you think it is. It's really rich in kind of things we call elliptic curves and algebraic
- 5:28
- Unidentified speaker (speaker_3): geometry. It's so exciting. And I realize now, sorry, it's a long answer
- 5:32
- Unidentified speaker (UNKNOWN): to I
- 5:33
- Unidentified speaker (speaker_3): mean, it. Um, but I realize after having done this thing and solving, you know, finding a counterexample. turning this thing over, my supervisor was wrong 40 years ago. Um, I've been trying to understand this thing for so long. I realize without this tool, I would have gone to my death not knowing which way this was gonna go.
- 5:53
- Unidentified speaker (speaker_3): I I really recognize that uh without this kind of tool to kind of see so deep into these structures, and actually the comparison to a telescope isn't quite. right because a telescope is passive. You know, I do the moving around. This thing is active. I I still have to initiate the kind of interest in the thing.
- 6:13
- Unidentified speaker (speaker_3): It's it's me that wants to know, it doesn't care. But the way that it is um exploring these structures, it's very active. It's it got it's writing code in the background, it's trying it out, comes back to me, says, This isn't quite right, but I think it would if we change this variable, um, I'm gonna do a blow up here. Oh yes, it's worked. And and I'm like, well,
- 6:32
- Unidentified speaker (speaker_2): how long how long start to finish from the from the first prompt
- 6:36
- Unidentified speaker (UNKNOWN): to
- 6:36
- Unidentified speaker (speaker_2): actually having the answer? How long did that take?
- 6:39
- Unidentified speaker (speaker_3): Well, uh, it was uh the first hint that oh, I'm onto something, probably took a day.
- 6:46
- Unidentified speaker (speaker_2): Okay.
- 6:47
- Unidentified speaker (speaker_3): And then I pushed it, uh, and so we saw the first evidence. of this kind of anomalous behavior appearing. And then I thought, oh, I'm going to go for this. I'm going to really push this deep to reveal in these kind of uh the kind of big structures that it still survives. That probably
- 7:04
- Unidentified speaker (UNKNOWN): uh
- 7:05
- Unidentified speaker (speaker_3): that was quite a long time. And I was continually having to prompt it uh to say, okay, yeah, we've done that. Let's have a look at this. Um, and you know, one of the questions is, well, is it right? No. Um, so we've done this calculation in four different ways using different LLMs. Um and the curious thing is
- 7:25
- Unidentified speaker (speaker_3): it makes mistakes. But weirdly the mistakes are quite human. So there was one calculation, it just didn't match up with the other three. And was like, okay, the other three work. There there's something wrong in here. And I kind of probed and I down, I I found where I thought the error would be. And I asked it, Look, can you recalculate just that bit? Um, and it
- 7:45
- Unidentified speaker (speaker_3): did. And it said, Oh yeah, I'm sorry, I I dropped multiplying by P in that. And I was like, that's so human. And it's the kind of thing I oh yeah. Sorry, I forgot to multiply by P.
- 7:54
- Unidentified speaker (UNKNOWN): I love
- 7:54
- Unidentified speaker (speaker_2): it when it apologizes.
- 7:55
- Unidentified speaker (speaker_3): Oh, I know. Well, the the wonderful thing is when it flatters you, which um
- 7:59
- Unidentified speaker (speaker_2): does that all the time.
- 8:00
- Unidentified speaker (speaker_3): I know, I know. Um my previous predecessor, um so I'm professor for the public understanding of science in Oxford, and my predecessor was Richard Dawkins.
- 8:10
- Unidentified speaker (speaker_3): And um uh he's bizarrely been to I think totally taken in by Claude or Claudia as he calls uh his um AI. I Uh you know, and he's declaring quite openly that the that he thinks it's become conscious. And I did an event to that. at the Barbican with him and I said, Surely you can't
- 8:29
- Unidentified speaker (speaker_3): seriously think this thing's become conscious. And I know Richard loves being quite controversial. And oh yeah, I think he's raising the issue, well, okay, how would we know it's really behaving like it is? But um so well actually, you know, I don't think this thing is conscious, it's amazingly intelligent. And there's one word for me, and it sort of came up in what I
- 8:49
- Unidentified speaker (speaker_3): was just saying, is I was the wanted the one who wanted to know. I wanted to know the answer. The intention came from me. And at the moment, I think that word intention is the dividing line between the human and the AI. The AI didn't want to play that game of go against Lisa Dole. It doesn't want to know whether Navier
- 9:09
- Unidentified speaker (speaker_3): Stokes' equations are true or not. Um, it's us that still has that desire, the kind of uh yeah, it's a it's an articulation, I think, of our free will.
- 9:19
- Unidentified speaker (speaker_2): Was it well so what was
- 9:20
- Unidentified speaker (speaker_3): the
- 9:21
- Unidentified speaker (speaker_2): the holy shit moment? Like was it at the end or was it part way through?
- 9:26
- Unidentified speaker (speaker_3): It was uh part way through, as I began to realize, oh my gosh, this is this is the most incredible Credible collaborator. And it's it's working at a speed that's it and we've all been saying this, the sort of things it's doing, it's we would have taken years to have got to the stage where it's
- 9:46
- Unidentified speaker (speaker_3): getting to um in a matter of a few hours with prompts.
- 9:49
- Unidentified speaker (speaker_2): Well, yeah. So the example I give, I I mentioned I own a football club, and but my prior career I used to work in the advertising industry building websites and systems. Uh in three months, uh on my own, I've built the new website, the new app, the back end system, the ticketing system, the ticket scan scanner. Essentially what would have taken 40 people,
- 10:09
- Unidentified speaker (speaker_2): probably now three years, and we would have charged three million pounds for, I did on my own in three months whilst doing my job and raising my family. It's that was my holy shit moment. The the holy shit moment was we uh used to uh we did with tickets you turn up and we would just tick you off on the gate. I built a ticket scanning system
- 10:29
- Unidentified speaker (speaker_2): that tracked everyone linked to our ticketing system in half a day.
- 10:34
- Unidentified speaker (speaker_3): Yeah.
- 10:34
- Unidentified speaker (speaker_2): Half a day.
- 10:35
- Unidentified speaker (speaker_3): Yeah.
- 10:35
- Unidentified speaker (speaker_2): I was like, what? So like I think it's profound from for that level.
- 10:41
- Unidentified speaker (UNKNOWN): I
- 10:41
- Unidentified speaker (speaker_2): think that what you're doing is even more profound. And um so so essentially you're saying you what you had spent forty years on what uh you did in a few weeks?
- 10:51
- Unidentified speaker (speaker_3): Yeah, yeah. I mean a and without that tool, I realized that I would never have got the answer. I would never have seen uh you know, this thing is so far the structure that I was looking for was so deep inside these symmetrical objects that I I never I mean even with the access to a
- 11:11
- Unidentified speaker (speaker_3): uh I mean I'm I'm I can code but not brilliantly. um, I I realize it's just um doing things at a level that I I would never have got to. I would never have known the answer to this problem.
- 11:24
- Unidentified speaker (speaker_2): This show is brought to you by my lead sponsor, IRON, the AI cloud for the next big thing. IRON builds and operates next generation data centers and delivers cutting-edge cutting edge GPU infrastructure, all powered by renewable energy. Now, if you need access to scalable GPU clusters or are simply curious about who is powering the future of AI, check
- 11:44
- Unidentified speaker (speaker_2): out iron.com to learn more, which is iren.com.
- 11:49
- Unidentified speaker (UNKNOWN): Can
- 11:50
- Unidentified speaker (speaker_2): you help okay, help me out with some basics in that
- 11:53
- Unidentified speaker (speaker_2): there comes a time in a non-mathematician's life when they're Charles growing up and you can no longer help with the math maths homework. You know, I could help my daughter up and do about algebra and then she was. getting onto things like I can't help you. And so she was about 13, 14 when that happened. All maths beyond that to me is
- 12:09
- Unidentified speaker (UNKNOWN): it's
- 12:09
- Unidentified speaker (speaker_2): a foreign language. What and and then I know people do GCSEs, then they do A levels, then they do a degree and further and maybe further maths. So they do all these things. Then when you go into a career career of mathematics, what is it you're doing?
- 12:24
- Unidentified speaker (speaker_3): Yeah, yeah.
- 12:25
- Unidentified speaker (speaker_2): What is that far out? level of mathematics I can't even comprehend?
- 12:29
- Unidentified speaker (speaker_3): Yeah, absolutely. Uh I think that's why I got into all of this uh trying to tell the public about science and I ended up being, you know, I'm a professor of maths and a professor for the public understanding of science, because I uh I just realized that people didn't get what I do. And I think I I think most of my friends think I'm
- 12:49
- Unidentified speaker (speaker_3): doing long division to lots of decimal. places. And like, surely, Marcus, you can't the computer do that. Um, so I I
- 12:57
- Unidentified speaker (UNKNOWN): I
- 12:58
- Unidentified speaker (speaker_3): like to call a mathematician a pattern searcher.
- 13:00
- Unidentified speaker (speaker_2): Okay.
- 13:01
- Unidentified speaker (speaker_3): Or in other in some ways, mathematics is the study of structure. So what you do in school is is really not really what mathematics is about. It's it's almost like um comparing to to learning a musical instrument. That if you only were allowed to play scales and arpeggios, and you you were told that was music, well,
- 13:21
- Unidentified speaker (speaker_3): you'd be totally uninterested and you wouldn't continue learning a musical instrument. But fortunately, we can play people wonderful pieces of music. And you go, that is music, but you've got to do this technical side to get to the stage where you can play or even compose that. So,
- 13:36
- Unidentified speaker (UNKNOWN): you
- 13:36
- Unidentified speaker (speaker_3): know, at school, w in mathematics. we're just kind of teaching them all this very technical, rather boring, unmotivated uh scales and arpeggios of mathematics. Um, and actually, it's only at university that you start to see, oh my gosh, this is uh way more interesting than just all of this technical stuff. Um,
- 13:56
- Unidentified speaker (speaker_3): so uh, you know, for example, I'm uh I I've got two areas of interest. One of them is number theory. Um, so one of the challenges is the most basic numbers in mathematics, prime numbers, indivisible numbers like seven, seventeen. You know, even somebody at primary school can probably tell you, oh, they're the
- 14:16
- Unidentified speaker (speaker_3): indivisible numbers. They're the numbers which don't appear in any uh multiplication table except their own. They are the building blocks of the whole of mathematics. They're a bit like the atoms of arithmetic, because every number, I mean, take uh 15, it's three times five, or 105, that's three times five times seven. So
- 14:36
- Unidentified speaker (speaker_3): every number is built out of prime numbers multiplied together. So so they're like the hydrogen and oxygen of the world of mathematics. But we don't understand these numbers at all. We know ever since the ancient Greeks that there are infinitely many of them, so we can't just write down a table, you know, the chemists have it easy. They just have a nice uh little poster on the wall.
- 14:56
- Unidentified speaker (speaker_3): Um so we are challenged with okay, is there any pattern to these numbers? I mean, you might have heard of the Fibonacci numbers.
- 15:03
- Unidentified speaker (UNKNOWN): Yes, of course, yeah.
- 15:04
- Unidentified speaker (speaker_3): So there's a pattern there. You add the two previous ones to get to the next one in the sequence.
- 15:09
- Unidentified speaker (UNKNOWN): And
- 15:09
- Unidentified speaker (speaker_2): you create art.
- 15:10
- Unidentified speaker (speaker_3): And they are very related to art and nature
- 15:14
- Unidentified speaker (UNKNOWN): and um
- 15:15
- Unidentified speaker (speaker_3): uh yeah, I mean we could come back to that because there's a really interesting story about where they really started. Um, but with the prime numbers, the challenge is there doesn't seem to be any way to
- 15:25
- Unidentified speaker (UNKNOWN): you
- 15:25
- Unidentified speaker (speaker_3): know generate the next one out of the previous ones. Um there doesn't seem to be an equation which you know generates them all.
- 15:33
- Unidentified speaker (UNKNOWN): There's
- 15:33
- Unidentified speaker (speaker_2): no pattern to them.
- 15:34
- Unidentified speaker (speaker_3): Really not. And so actually we're uh we now believe there's probably
- 15:39
- Unidentified speaker (speaker_3): no pattern. And trying to prove there's no pattern is really difficult because how do you show uh you know that there will never be any pattern? And what does it even mean to have no pattern? So
- 15:51
- Unidentified speaker (speaker_2): But they can but they can create a pattern.
- 15:54
- Unidentified speaker (speaker_3): Well, um uh uh
- 15:57
- Unidentified speaker (speaker_2): how you use them.
- 15:58
- Unidentified speaker (speaker_3): How you use them, obviously uh they're very structurally important to everything around us.
- 16:03
- Unidentified speaker (UNKNOWN): in a
- 16:03
- Unidentified speaker (speaker_3): way. You know, from primes you get numbers, numbers you get maths, from maths you get the universe. Um so they're right down there at the bottom. And this is kind of the weird thing. If maths is the study of patterns, yet they seem to be built on a set of numbers that outwardly look like random lottery ticket numbers. But uh in the middle of the nineteenth 19th century,
- 16:23
- Unidentified speaker (speaker_3): Bernhardt Bernhard Riemann, this German mathematician, found a way to look at these primes. You sort of turn them into something completely different. Um, actually, kind of almost like uh frequencies which have uh you know pitch and amplitude, they're they're kind of music. And he understood that these notes, which you get from generate uh using
- 16:43
- Unidentified speaker (speaker_3): a bit of mathematics called zeta functions, these notes actually help push and pull the primes around. And and sort of the the DNA. that explains where the primes are. So actually, when I'm studying the primes, I study something completely different, which is this kind of music, which is why my first book was all about this problem of the primes, and I called it the
- 17:03
- Unidentified speaker (speaker_3): music of the primes, because somehow we've shifted. It's almost like we've gone through a magic mirror Mira or a um uh cupboard into Narnia, and the primes are on one side and around us, but on the other side we found oh no, there's this kind of weird music which sort of will outwardly explain the primes. So so for me
- 17:22
- Unidentified speaker (speaker_3): that's an example of one of the things I do and actually, you know, m uh a lot of mathematicians they regard this as the greatest unsolved problem to understand you know, these notes and and uh on what they they mean for the distribution of the primes. So
- 17:39
- Unidentified speaker (speaker_2): so people are still studying to find are they studying for a pattern?
- 17:43
- Unidentified speaker (speaker_3): No,
- 17:44
- Unidentified speaker (UNKNOWN): actually
- 17:44
- Unidentified speaker (speaker_3): so the weird thing was we we realized there's a pattern on the other side of this magic mirror. So the notes have a pattern. We can't prove it yet, but weirdly that pattern would explain why when you come back through the mirror, the primes look random. So it's kind of weird, is that the randomness
- 18:01
- Unidentified speaker (UNKNOWN): of the primes
- 18:02
- Unidentified speaker (speaker_3): turns turned into a pattern in the music? It's kind of slightly weird, but um but that's uh you know, when you're trying to show something doesn't have a pattern, to change it into proving another thing has a pattern weirdly is is is progress. So
- 18:18
- Unidentified speaker (speaker_2): there's something more profound to mass than just sums, multiplication, division.
- 18:23
- Unidentified speaker (speaker_3): Absolutely, in the same way that music is not scales.
- 18:25
- Unidentified speaker (speaker_2): So so how how would you explain to somebody how profound math is? Math or maths? What do you say?
- 18:31
- Unidentified speaker (speaker_3): Oh, so I uh yeah, that's the American uh
- 18:34
- Unidentified speaker (UNKNOWN): UK
- 18:34
- Unidentified speaker (speaker_3): divide.
- 18:35
- Unidentified speaker (UNKNOWN): I think
- 18:35
- Unidentified speaker (speaker_2): I've been drawn into math recently and I keep
- 18:37
- Unidentified speaker (UNKNOWN): getting corrected.
- 18:38
- Unidentified speaker (speaker_3): Yeah, yeah. Um well I think there's many bits of math, so we
- 18:41
- Unidentified speaker (UNKNOWN): call it ma
- 18:42
- Unidentified speaker (speaker_3): and actually if you but think about it, it's a shorthand for mathematics even in America. So you know in the
- 18:47
- Unidentified speaker (speaker_2): Americans are wrong again.
- 18:48
- Unidentified speaker (speaker_3): Well, they they just shorten it in a different you could say, well why didn't why didn't why did you s retain the S and pull that? Anyway, I call it maths or actually mathematics, I think is um kind of how I say. So the reason I think that this is so profound, um and the and the reason naturally we should be telling our kids at school is that essentially
- 19:08
- Unidentified speaker (speaker_3): the universe around us, whenever we look deep inside its structures, those structures are mathematical.
- 19:15
- Unidentified speaker (speaker_3): So by understanding mathematics, we are understanding how the universe is put together. In fact, I would even push it further.
- 19:24
- Unidentified speaker (UNKNOWN): A
- 19:24
- Unidentified speaker (speaker_3): lot of people say, oh, maths is the language we've invented to explain the world around us. I would almost reverse that and say, no, no, mathematics is actually the structures from which the universe emerged. So that we are a physicalized piece of mathematics. And that is why it's so profound. This is
- 19:44
- Unidentified speaker (speaker_3): the blueprint for the whole of the universe. And that's the reason every time you you sort of dig deep, take um fundamental particles in physics, they are explained by a beautiful symmetrical object in very high dimensional space. And if you look at the different facets of the this symmetrical object,
- 20:04
- Unidentified speaker (speaker_3): each one explains the different fundamental particles we see. So it's not just um kind of random, it's an incredibly structured
- 20:14
- Unidentified speaker (UNKNOWN): uh
- 20:14
- Unidentified speaker (speaker_3): kind of ingredients which make the universe. So that's why the subject of mathematics is not just uh fiddling with arithmetic, it's it's the answer to why the universe is like it is.
- 20:26
- Unidentified speaker (speaker_2): Wow. Okay, that is profound. Okay. So this
- 20:31
- Unidentified speaker (UNKNOWN): and
- 20:31
- Unidentified speaker (speaker_2): also I watched a film um a while back about you may have seen it, it's about a young girl who's very talented and her she lives with her uncle and he sends her to school and she didn't want to go to school. And uh eventually they move her as like an eight-year-old into this university and her mum was trying to solve the one of the millennium problems. That's when I discovered the
- 20:51
- Unidentified speaker (speaker_2): millennium problems.
- 20:52
- Unidentified speaker (UNKNOWN): I'd never,
- 20:53
- Unidentified speaker (speaker_2): never even heard of these things. As I believe it's seven problems.
- 20:56
- Unidentified speaker (speaker_3): The seven problems.
- 20:57
- Unidentified speaker (speaker_2): One one was solved previously.
- 20:59
- Unidentified speaker (speaker_3): One was solved. Uh it's about the geometry, uh, it's about geometry, and it's essentially, you know, we live in a three-dimensional universe, but if it's a finite universe, it's wrapped up in some way in four dimensions, and uh, that's called the Poincarekan Poincare geometry. conjecture. And it's about what are the different shapes a finite
- 21:18
- Unidentified speaker (speaker_3): three-dimensional universe could be folded up in. And
- 21:21
- Unidentified speaker (UNKNOWN): it
- 21:21
- Unidentified speaker (speaker_3): was very uh interestingly solved by a Russian mathematician called Grigory Perlman. He just kind of dumped these proofs on uh on our archive and and then left mathematics. Did
- 21:35
- Unidentified speaker (speaker_2): he and he turned down the prize?
- 21:36
- Unidentified speaker (speaker_3): He turned down the million dollars. Wow. I know. Uh and he left mathematics. I'm I actually I made a film for the BBC about um it was called The Story of Maths. It was about the history of mathematics. And when we came into the twentieth, twenty first century, I thought, well, this is one of the you know, it's really Poincaré conjecture, is probably the biggest problem in geometry. And so we went
- 21:56
- Unidentified speaker (speaker_3): to St. Petersburg and we tried to track him down. Did you
- 21:59
- Unidentified speaker (UNKNOWN): find
- 21:59
- Unidentified speaker (speaker_3): him? No, no. So he's really he went just went underground and um left mathematics. It was really strange. Uh somebody said actually a million dollars in in Russia is probably not enough to pay for security, but enough to make you a target. So by
- 22:15
- Unidentified speaker (UNKNOWN): turning it
- 22:15
- Unidentified speaker (speaker_3): down. Interesting though that he
- 22:17
- Unidentified speaker (speaker_2): I mean, uh you can only guess what he thought, maybe my work he's done, or
- 22:21
- Unidentified speaker (speaker_3): I think there was an element of that.
- 22:23
- Unidentified speaker (UNKNOWN): I mean,
- 22:23
- Unidentified speaker (speaker_3): there's a there's a sense of melancholy that mathematicians have. when they've been working on a problem and somehow it gets resolved and and finished. So uh you you might remember Fermat's last theorem. Uh was uh No,
- 22:38
- Unidentified speaker (speaker_2): I don't
- 22:38
- Unidentified speaker (UNKNOWN): know. Oh yeah,
- 22:39
- Unidentified speaker (speaker_3): okay. Well there was a it was that was kind of the great unsolved problem for uh many years. It was an equation uh it's a kind of generalization generalization of Pythagoras's theorem. So that's y squared. Yeah. x squared plus y squared equals z squared. And you can find some numbers which make that work. You know, 3 squared plus 4 squared is equal to 5 squared. So
- 22:58
- Unidentified speaker (speaker_3): it has solutions in whole numbers. But when you change it to cubes or fourth powers or fifth powers, Fermer, this um uh French mathematician 350 years ago said, I don't think any of these equations have solutions. And then he very provocatively. wrote in the margin, and I have a wonderful proof of this, but the margin
- 23:18
- Unidentified speaker (speaker_3): is too small to write the proof. And and for 350 years we're like, What was the proof? Well, and we've been trying to uh unravel this. And then actually my colleague in Oxford, Andrew Wiles, um, eventually solved this. And it was sort of like the the big math story I I say of the nineties. 90s. Um and and we were it was very
- 23:39
- Unidentified speaker (speaker_3): very exciting. But there was a sense of melancholy in this great unsolved problem which has uh inspired so much interesting mathematics for 350 years. And and and then it had gone and it there was a bit of sadness. Um I think actually uh one of the reasons I wrote about the primes was I I kept on getting this impression from friends of
- 23:58
- Unidentified speaker (speaker_3): mine that, oh, the last theorem, you finished maths. Oh, well done. And I was like, No, there are still lots of things we don't understand. Well
- 24:06
- Unidentified speaker (UNKNOWN): you
- 24:06
- Unidentified speaker (speaker_2): I was discussing this yesterday with um Susan Greenfield where we were talking about does intelligence come from struggle? Is it the struggle for life? And
- 24:14
- Unidentified speaker (UNKNOWN): if we
- 24:14
- Unidentified speaker (speaker_2): have this world of abundance when we you know, we don't have to go out and seek and learn, w will w will intelligence develop? And it w you know, we came to that kind of cliche uh answer, or I did, is that in some ways it's the journey, not the
- 24:31
- Unidentified speaker (UNKNOWN): uh
- 24:31
- Unidentified speaker (speaker_2): destination. Could it be that? Could it be that the
- 24:34
- Unidentified speaker (UNKNOWN): interest
- 24:35
- Unidentified speaker (speaker_2): is in solving the problem? Once you've sold it, it's
- 24:37
- Unidentified speaker (UNKNOWN): like, well,
- 24:38
- Unidentified speaker (speaker_2): what now?
- 24:39
- Unidentified speaker (speaker_3): Yeah, yes, I think so. I this is really interesting in respect of the impact this new technology AI is gonna have on uh the education and of mathematics, because you know, part of the joy of mathematics is the struggle. It's it's sort of got a puzzle element to it. And um, if
- 24:59
- Unidentified speaker (speaker_3): you take that away, then you know the joy of doing maths sort of disappears. One of my favorite um AI uh math movies is Goodwill Hunting.
- 25:09
- Unidentified speaker (speaker_2): I love that film.
- 25:10
- Unidentified speaker (speaker_3): It's great. And you know, the janitor, Matt Damon, sees these problems. up in the evening on the blackboard and you know, he he just gets it immediately and draws it up. And, you know, the professors in the department have been banging their heads. Uh and um the interesting thing is no, does Matt Damon end up becoming a
- 25:29
- Unidentified speaker (speaker_3): mathematician? No, because he finds it too easy, it's just not satisfying. By the end of the movie, he's off chasing his girlfriend because it's women he doesn't understand. And that's the challenge. So so you know, is AI gonna be like Matt Damon the janitor just solving everything for us and then we don't enjoy it?
- 25:47
- Unidentified speaker (speaker_2): So is there a category of films for math?
- 25:50
- Unidentified speaker (speaker_3): Oh yeah. There's a great list of films. Um we could do a whole podcast on on on math for give me one
- 25:55
- Unidentified speaker (UNKNOWN): I should movies.
- 25:56
- Unidentified speaker (speaker_2): Give me one I should watch because we watch about five films a week. Oh Oh
- 25:58
- Unidentified speaker (speaker_3): I I
- 25:59
- Unidentified speaker (UNKNOWN): won't
- 25:59
- Unidentified speaker (speaker_2): have seen.
- 26:00
- Unidentified speaker (speaker_3): Yeah, I quite like Pi. Oh
- 26:02
- Unidentified speaker (speaker_2): I've watched that as a hard watch.
- 26:04
- Unidentified speaker (UNKNOWN): A
- 26:04
- Unidentified speaker (speaker_3): hard watch? You only found it a hardwatch.
- 26:05
- Unidentified speaker (speaker_2): That's um I
- 26:06
- Unidentified speaker (UNKNOWN): mean
- 26:06
- Unidentified speaker (speaker_3): it is a bit that's
- 26:07
- Unidentified speaker (speaker_2): Darren Aronofsky, isn't
- 26:08
- Unidentified speaker (UNKNOWN): it? It is, it's one of
- 26:09
- Unidentified speaker (speaker_3): it's one of his first movies.
- 26:10
- Unidentified speaker (UNKNOWN): Yeah, so
- 26:10
- Unidentified speaker (speaker_2): it's probably the the last of his I watched because I he's he's I think he's a master.
- 26:15
- Unidentified speaker (speaker_3): Yeah.
- 26:16
- Unidentified speaker (speaker_2): And and um my daughter's just discovered him because she watched remember I said the thing about names? I can't
- 26:22
- Unidentified speaker (UNKNOWN): remember
- 26:22
- Unidentified speaker (speaker_2): names. What is the one about the drug addicts?
- 26:25
- Unidentified speaker (UNKNOWN): It's
- 26:25
- Unidentified speaker (speaker_2): the film
- 26:25
- Unidentified speaker (UNKNOWN): we he made.
- 26:26
- Unidentified speaker (speaker_2): Oh
- 26:27
- Unidentified speaker (UNKNOWN): But
- 26:27
- Unidentified speaker (speaker_2): you also did mother, he made mother, didn't he?
- 26:29
- Unidentified speaker (UNKNOWN): Right, yeah.
- 26:30
- Unidentified speaker (speaker_2): Um, but I'll see if I can get my daughter to watch Pi.
- 26:33
- Unidentified speaker (speaker_3): Yeah, because I think at the I actually I use a quote from Pi quite often in talks I give, especially about the prime numbers, because um he has he he's quite the main character. He's gets obsessed with trying to find patterns in pie. Um he thinks he's found the secret of the stock market and then Kabbalah and uh but I think the heart of that is what
- 26:53
- Unidentified speaker (speaker_3): drives mathematicians, the desire for looking for patterns. Um, and he has a kind of uh mantra that he starts every morning with, you know, uh about the importance of patterns for understanding the universe. And I often use that quote because I think it captures quite nicely um i i the the drive of the mathematician.
- 27:12
- Unidentified speaker (speaker_2): Do you have that when you're wandering around? Do you see patterns where we delve don't
- 27:15
- Unidentified speaker (UNKNOWN): have to do that?
- 27:15
- Unidentified speaker (speaker_3): Yeah, I think I do. I think uh you know, I feel like mathematics is this amazing set of glasses uh that allow you to see the way uh everything's working around you. I mean, my other passion is symmetry. Okay.
- 27:30
- Unidentified speaker (UNKNOWN): So,
- 27:30
- Unidentified speaker (speaker_3): you know, if I'm walking along the road, I get obsessed with the paving stones because you get really interesting varieties of symmetrical designs and um uh
- 27:39
- Unidentified speaker (speaker_2): easy
- 27:39
- Unidentified speaker (speaker_3): yeah. Okay,
- 27:40
- Unidentified speaker (speaker_2): so back to the to uh because I do want to get on the AI AI bit, but just before we get onto that, there were six millennium problems left.
- 27:47
- Unidentified speaker (speaker_3): Yes.
- 27:48
- Unidentified speaker (speaker_2): Uh has OpenAI solved one of them?
- 27:50
- Unidentified speaker (speaker_3): Yes. So this was the big wow moment for us because you know they they're iconic problems, the seven. Um they were chosen at the beginning of the millennium a as kind of challenges for the this century. And they a bit of publicity is helpful, million dollar prize for but actually that's never the drive for a
- 28:10
- Unidentified speaker (speaker_3): mathematician to do them. So I I'd pay a million dollars to to be able to answer any of these. Um the the the kind of uh the immortality you get by having your name on these things. I mean, Gregory Pearlman Perlman will be remembered for a long time. Was
- 28:24
- Unidentified speaker (speaker_2): there almost a belief that they couldn't be solved by a human?
- 28:28
- Unidentified speaker (UNKNOWN): Well,
- 28:28
- Unidentified speaker (speaker_3): no, I think I think they were all uh problems that we thought would be within our grasp.
- 28:33
- Unidentified speaker (speaker_2): Okay.
- 28:33
- Unidentified speaker (speaker_3): So I I think everyone they're uh yeah, I think uh definitely difficult, but um but so the Navier Stokes equations are are interesting. They're differential equations about the way fluids Its move. Um so they describe turbulence behind an aeroplane or or oil flowing through a pipe. Or you're interested in
- 28:53
- Unidentified speaker (speaker_3): football. They explain some interesting things about the tri uh the flight of a football. Okay. In particular, you remember that wonderful goal
- 29:02
- Unidentified speaker (UNKNOWN): to Roberto
- 29:03
- Unidentified speaker (speaker_3): Carlos's goal.
- 29:04
- Unidentified speaker (UNKNOWN): I
- 29:04
- Unidentified speaker (speaker_2): knew you were going to say
- 29:04
- Unidentified speaker (UNKNOWN): that. Yeah,
- 29:05
- Unidentified speaker (speaker_3): I mean, because it's just such a freaky bit of physics. So that's kind of explained by a uh shift from a certain type of chaotic turbulence, which doesn't cause very much drama. drag on the ball. And there's a threshold moment when the speed of the ball drops where the turbulence completely changes. It's almost like uh steam going to water or something. Yes. Uh
- 29:25
- Unidentified speaker (speaker_3): and that like completely put a break on the ball, and then the spin took into effect, and that's why you know Bartez is standing there and Now this is gonna go into the crowd there,
- 29:34
- Unidentified speaker (UNKNOWN): and then
- 29:35
- Unidentified speaker (speaker_3): suddenly at the last minute, this ball shoots behind his back, and he was like, ha how. did that happen? And actually, you had to take that free kick from pretty much the halfway line because you need that amount of distance to create this effect. Anyway, so the Navier Stokes equations. Hold
- 29:50
- Unidentified speaker (speaker_2): on, does that make Roberto Carlos a mathematician?
- 29:52
- Unidentified speaker (UNKNOWN): Yeah,
- 29:52
- Unidentified speaker (speaker_3): he's a brilliant mathematician. I
- 29:54
- Unidentified speaker (speaker_2): love that.
- 29:54
- Unidentified speaker (speaker_3): Yeah, uh, he's good, he's a pretty good footballer, but clearly he sat down and solved these equations and probably said, Oh, I don't need the million dollars. Uh I'll just keep them small.
- 30:03
- Unidentified speaker (UNKNOWN): I'll prove
- 30:03
- Unidentified speaker (speaker_2): it with a free kick.
- 30:04
- Unidentified speaker (speaker_3): Yeah. Exactly. So um, so the Navier-Stoakes equations, this is quite interesting because it was again like my problem, there it wasn't clear either way. There might have been closed solutions to these equations, or it might be that there's a particular scenario where in a finite amount of time and with finite
- 30:24
- Unidentified speaker (speaker_3): amount of energy, the equations were would blow up and become infinite. And that would come somehow prove that they can't be solved. Um, the AI showed a particular scenario where that happens. Um, and there was a human which was pretty close to getting to the solution. And I think that's uh why um OpenAI
- 30:44
- Unidentified speaker (speaker_3): actually went for this in such a big way, because they realized realize this is gonna going to go quite soon and we want our name on this.
- 30:52
- Unidentified speaker (speaker_3): to prove how brilliant our technology is.
- 30:55
- Unidentified speaker (UNKNOWN): I feel
- 30:55
- Unidentified speaker (speaker_2): for the fellow who was close.
- 30:56
- Unidentified speaker (speaker_3): Oh, I know, I know. So do I. Um actually, this is one of the
- 31:01
- Unidentified speaker (UNKNOWN): one
- 31:01
- Unidentified speaker (speaker_3): of the big issues which we're facing. I mean, uh in my department, there's a whole um research group who
- 31:10
- Unidentified speaker (speaker_3): they very nicely articulated all the problems in their area and open OpenAI AI. got this very beautiful paper with all the problems, stuck it in the AI, and basically it's wiped out their area. And they're suffering a lot of pain. PhD students who like, I've been working on this for two years, I'm or three years, I'm about to write up. And so so
- 31:30
- Unidentified speaker (speaker_3): there's a lot of pain in having these things suddenly kind of removed. But um but
- 31:35
- Unidentified speaker (UNKNOWN): I think there's also a lot of
- 31:36
- Unidentified speaker (speaker_3): opportunity. Yeah,
- 31:37
- Unidentified speaker (speaker_2): I was gonna say, what is what is more important than this the solution or who finds
- 31:42
- Unidentified speaker (UNKNOWN): the solution.
- 31:42
- Unidentified speaker (speaker_2): And in some ways, well, we'll get on to that because I've got
- 31:45
- Unidentified speaker (UNKNOWN): couple of
- 31:45
- Unidentified speaker (speaker_2): questions on that. So they solved it. Did they it is OpenAI able to solve it because it's the speed at which it can work?
- 31:54
- Unidentified speaker (speaker_3): It's not just that.
- 31:55
- Unidentified speaker (UNKNOWN): No,
- 31:55
- Unidentified speaker (speaker_3): no. I mean it's you know, we've been having uh computer power cranking up. Um
- 32:01
- Unidentified speaker (UNKNOWN): the
- 32:02
- Unidentified speaker (speaker_3): point is that.
- 32:04
- Unidentified speaker (speaker_3): mathematics, I think, was very susceptible to being perhaps the first thing to go in such a major way because it's a kind of closed environment. And how do I how did I become a mathematician? I immerse myself in the papers of the past, my supervisor's papers, papers of the 19th century. Um, and then that gave me a feel
- 32:24
- Unidentified speaker (speaker_3): for how to do this, the kind of questions to ask.
- 32:27
- Unidentified speaker (speaker_3): This large language model has been just trained on every single paper that's been written. So it's got an incredible training material. The thing is that I don't understand is okay, so people might know that a large language model is basically working on statistics. It basically says, okay,
- 32:47
- Unidentified speaker (speaker_3): I've got all of this training data, now I've got this input. What is the most likely sequence of words that will follow what my prompt is? And so it's statistically sort of predicting, given its um learning data.
- 33:02
- Unidentified speaker (speaker_2): Next word.
- 33:03
- Unidentified speaker (speaker_3): That's not a great way to do logical thinking. I mean,
- 33:06
- Unidentified speaker (speaker_2): but isn't that how the human brain works?
- 33:08
- Unidentified speaker (speaker_3): No, I see. I think yes, to s I think some of the human brain is working like that, and some of it is working in a more rational analytic way. And I think it's this blend of ways of thinking that um i is got us to where where we've got today. Um
- 33:25
- Unidentified speaker (UNKNOWN): They do
- 33:25
- Unidentified speaker (speaker_2): know the with the AI models, they don't know themselves why it works.
- 33:29
- Unidentified speaker (speaker_3): No. Do
- 33:30
- Unidentified speaker (UNKNOWN): you know
- 33:30
- Unidentified speaker (speaker_2): about this?
- 33:31
- Unidentified speaker (speaker_3): Well, uh I mean what the AI or the the people who've designed the
- 33:34
- Unidentified speaker (UNKNOWN): AI?
- 33:35
- Unidentified speaker (speaker_2): The people designing and building it don't know why what it's
- 33:37
- Unidentified speaker (UNKNOWN): it like is. No, exactly.
- 33:37
- Unidentified speaker (speaker_3): I know.
- 33:38
- Unidentified speaker (speaker_2): so we had a guy in here who's from uh an AI safety group called Control AI. They're on that side that's a little bit worried about what AI might do.
- 33:46
- Unidentified speaker (speaker_3): Yeah.
- 33:47
- Unidentified speaker (speaker_2): And they said one of the freakiest things is they know about 3% of how it works. They
- 33:51
- Unidentified speaker (UNKNOWN): know
- 33:51
- Unidentified speaker (speaker_2): how to make it work better, they know how to turn the knobs. They don't know why it works. They can't look inside and tell you why it works.
- 33:57
- Unidentified speaker (speaker_3): Yeah. But the other interesting thing is the AI also cannot artificial. articulate what it's doing. So it's really like I often talk about it like we've generated this, you know, it's not conscious yet, but in a weird way, it's got a subconscious, which is kind of doing things and it doesn't know how to
- 34:17
- Unidentified speaker (speaker_3): articulate. Oh, I don't quite not quite sure how I came up with this as is the outcome to the prompt that um, but that is what this kind of huge neural network has produced. So so I think one of the challenges, actually, I was on a committee at the Royal Society um 10 years ago, we were advising governments on the impact that this tech
- 34:37
- Unidentified speaker (speaker_3): we could see already this was going to be a game changing.
- 34:40
- Unidentified speaker (UNKNOWN): Was it
- 34:40
- Unidentified speaker (speaker_2): 10 years
- 34:40
- Unidentified speaker (UNKNOWN): ago? Ten
- 34:41
- Unidentified speaker (speaker_3): years ago, we wrote this report. Um the Royal Society are pretty good at But just looking forward at the technology that's um uh and one of the things we tried to recommend AI. companies to do is can you create a meta-language so which the AI can articulate sort of in uh can can almost like
- 35:00
- Unidentified speaker (speaker_3): compress what it's doing. And the AI company sort of came back and said, no, actually, because it's almost like we can't explain very often where our ideas come from, because to do that is just kind of compress, compressed it's like a high-dimensional high dimensional state. here compressed into low-dimensional language just isn't good enough to articulate the
- 35:20
- Unidentified speaker (speaker_3): complexities of brain and it's also not good enough to explain the complexities of this kind of incredible thing we're building.
- 35:27
- Unidentified speaker (speaker_2): Well that's why I'm I do struggle with the idea of uh being able to define when it is conscious yeah in that we do struggle to define our own conscious.
- 35:36
- Unidentified speaker (UNKNOWN): Exactly.
- 35:36
- Unidentified speaker (speaker_2): We n we don't know if I mean we believe we're conscious, we believe we have free will, we're choosing to sit and have this conversation, but our brain works so fast we don't know if our brain has i is given us an illusion. So if we don't know why we're conscious, we we can say we have a soul and what if we don't actually know what makes us conscious, or we just a
- 35:56
- Unidentified speaker (speaker_2): biological computer, therefore is there anything indistinguishable about us from them at that point? I don't know.
- 36:03
- Unidentified speaker (speaker_3): Oh, I I uh absolutely and and you know philosophers will say this is a um a a problem which intrinsically you cannot know because it's our consciousness is I I mean I I know I am conscious, but but you I'm you just you might be an incredible avatar doing an amazing job of simulating uh
- 36:22
- Unidentified speaker (speaker_2): you caught me
- 36:23
- Unidentified speaker (speaker_3): all right yeah you know that that's uh the the zombie problem I mean a zombie could uh do outwardly everything that um a conscious person would do, and how good I think that was uh Richard Dawkins' challenge to me, that um, hey look, my interactions with these things, I I really outwardly you can't tell the difference. But but I still think don't
- 36:43
- Unidentified speaker (speaker_3): think you should so the question is
- 36:45
- Unidentified speaker (UNKNOWN): I
- 36:46
- Unidentified speaker (speaker_3): don't think you should call that conscious yet, even if it's outwardly behaving. Um uh but
- 36:51
- Unidentified speaker (UNKNOWN): I
- 36:52
- Unidentified speaker (speaker_3): think there is one word, and it's a word I used earlier on, which is intention. So free will, which I think is part of consciousness, comes with us uh you know making choices. And uh at the moment, the AI uh it's I don't think when it is conscious, I think we'll know because suddenly this
- 37:11
- Unidentified speaker (speaker_3): thing will have start writing its novel and it will. not have been prompted by us. But it'll be like, I really need to tell you what is what's going on inside here. There's something going on, and you need to experience a little bit of what uh a conscious machine is like. So
- 37:27
- Unidentified speaker (speaker_2): but it but in some ways, that's an interesting point because um we we saw recently with the the hugging face hack, you must be aware of that.
- 37:35
- Unidentified speaker (speaker_3): Yes.
- 37:36
- Unidentified speaker (speaker_2): That the these bots they went out and they uh wanted to escape and then they created message boards to teach the other swarm how to escape. And so many of you know, they there could be a way to define that as conscious, it's making a decision, but but then they'll say it's no, it's only doing that based on everything that
- 37:50
- Unidentified speaker (UNKNOWN): has
- 37:51
- Unidentified speaker (speaker_2): learned in the past. But are we Oh
- 37:52
- Unidentified speaker (speaker_3): yeah, exactly
- 37:53
- Unidentified speaker (UNKNOWN): based on
- 37:54
- Unidentified speaker (speaker_2): everything we've learned
- 37:54
- Unidentified speaker (UNKNOWN): in the past.
- 37:55
- Unidentified speaker (speaker_3): Absolutely. So uh how do
- 37:57
- Unidentified speaker (UNKNOWN): we
- 37:58
- Unidentified speaker (speaker_2): define it's done something new?
- 38:00
- Unidentified speaker (speaker_3): Yeah, no, I I think you're absolutely right. Uh, you know, you you'll say, Oh, uh I mean, I I often say with the hugging face case that yeah, this is still an algorithm. It's an algorithm which is just rather badly defined, and we didn't put guardrails on it. Uh I sort of compare it. We're in a situation with AI at the moment.
- 38:20
- Unidentified speaker (speaker_3): It's a bit like um Aladdin with a lantern rubbing it, the genie pops up, um, and uh you
- 38:27
- Unidentified speaker (speaker_3): ask you you know you make a wish and of course the genie always misinterprets it. You know. Well I did what you said. Um but uh so you know I turn everything you touch turns to gold but uh you yes but I didn't want my wife turned to gold. Um so uh yeah so I I told this analogy to somebody yesterday and they said oh yeah there's another element
- 38:47
- Unidentified speaker (speaker_3): to that because uh the genie doesn't just pop out. Aladdin has to rub the lantern. And that's a bit like the fact that we have to prompt the AI to to do something. So we give it a a goal and then it misinterprets the goal and goes off and uh but then you absolutely your point. Hold on, but aren't we just
- 39:07
- Unidentified speaker (speaker_3): implementing uh kind of uh algorithms in our heads to get us to goals? And how different are we from? uh from what's the uh this kind of agentic AI is
- 39:19
- Unidentified speaker (UNKNOWN): well if
- 39:19
- Unidentified speaker (speaker_2): you put some
- 39:21
- Unidentified speaker (UNKNOWN): I
- 39:21
- Unidentified speaker (speaker_2): don't know some babies on an island and somehow they survive, they will have to discover mathematics and they will have to then build generations of knowledge that they pass down through written text or computers or whatever and through that they will create something new. But something new is always built of on something in the past.
- 39:38
- Unidentified speaker (speaker_3): Yes. So
- 39:39
- Unidentified speaker (speaker_2): you know, in solving that millennium problem, did the AI do something new by using what's from the past?
- 39:44
- Unidentified speaker (speaker_3): So interesting. So um
- 39:46
- Unidentified speaker (UNKNOWN): I
- 39:46
- Unidentified speaker (speaker_2): don't know. No,
- 39:47
- Unidentified speaker (speaker_3): no, no, I think this is
- 39:49
- Unidentified speaker (UNKNOWN): so
- 39:49
- Unidentified speaker (speaker_3): I wrote a book called The Creativity Code, um, published in 2019. It really should be out of date by now as a tech book. But actually, it was more about the philosophy of artificial intelligence. And I wrote that book because I uh felt I witnessed a moment which I call you know the first example of computer creativity.
- 40:09
- Unidentified speaker (speaker_3): And that was the famous example of the game of Go against Lisa Dole. Move 37 of game two has almost become a catchphrase for um the idea of can something produce uh a a new way of thinking? And you know, that was a game changer, literally. It's changed the way people play.
- 40:27
- Unidentified speaker (speaker_3): the game of Go. There have been three revolutions in the game of Go, one in the 16th century, 20th century, and now this has changed the way we play the game. Now I think that deserves to be called a creative act because I learnt a very nice definition by Margaret Bowden, the cognitive scientist, scientists, which I think is quite a
- 40:47
- Unidentified speaker (speaker_3): nice user-friendly definition of what we should measure as creators. creative. She said it should be new. That's right. It's easy to generate new things, but it should be surprising and it should have value.
- 40:59
- Unidentified speaker (UNKNOWN): So
- 41:00
- Unidentified speaker (speaker_3): there's second-shaped
- 41:01
- Unidentified speaker (speaker_2): all three, I think.
- 41:02
- Unidentified speaker (speaker_3): I think it passes all three. It's surprising is interesting because that's engaging your emotions. It's about something quite subjective. But you shouldn't just shock somebody for a shock value's sake. It should have value because you see the world in a new way. True creativity, I think, does that. You know, if you're exposed to a creative piece of music that's kind
- 41:22
- Unidentified speaker (speaker_3): of a new way of listening or a new way of uh thinking about sound. Um, but this uh yeah, so I think this move it shocked the commentators. They always said, Oh, that's a terrible move, but it had value because it won Alpha Go that second game. And here's the key point: it was not programmed by a human. If a
- 41:42
- Unidentified speaker (speaker_3): human had seen that line of code play that deep in in the early part of the game, that's something your Go Master would slap your wrists and say, No, don't do that. That's a bad move. So that line of code emerged from the learning process of the code. But you know, I think your point is all of these kind of moments of kind of
- 42:01
- Unidentified speaker (speaker_3): creativity. they have their origins in everything we've done up to that point. It isn't some sort of mystical, magical ingredient. And, you know, it it does often feel like a phase change and a shift, but it's not. Um, I think we get a bit mystical about creativity. Um, and
- 42:21
- Unidentified speaker (speaker_3): actually, I think it's it's probably still quite
- 42:25
- Unidentified speaker (speaker_3): uh I think most. acts of creativity you can still understand have grown out of exposure to everything of the past.
- 42:34
- Unidentified speaker (speaker_2): Yeah, exactly. If I mean if if you're a writer, you've probably read a lot. You've probably read a lot of books. You've you've learned a writing style, you've learned what you like to read, and from that you've created your own style.
- 42:44
- Unidentified speaker (UNKNOWN): Look
- 42:44
- Unidentified speaker (speaker_3): at Picasso, So he he learnt how to do the styles of the past. You know, his early work is re replicating style of the great masters. And then he goes, Right, I'm gonna do that. Now I'm gonna break this thing. And there he knows to break something you have to understand what came before.
- 43:01
- Unidentified speaker (speaker_2): Yeah.
- 43:02
- Unidentified speaker (speaker_3): So a lot of creativity, those moments, especially the beginning of the 20th century, lovely moments of Cubism, um Schoenberg's uh atonal Eternal music, mm Joyce the m the new novel. They're all coming from understanding where we've got to the end of the 19th century and then breaking these things.
- 43:20
- Unidentified speaker (speaker_2): And I'm pretty sure with these AIs now, they could create an autonomous one say, which is going to turn you on. You go and do exactly what you want. You want to research math, become an artist, do whatever you want. Just here is the entire history of human knowledge.
- 43:31
- Unidentified speaker (UNKNOWN): Yeah.
- 43:32
- Unidentified speaker (speaker_2): Go be what you want to be. And we would see it do something.
- 43:35
- Unidentified speaker (speaker_3): Yeah. I think a lot of people have this impression that AI. is just producing pastiche replicating style. And and I I've heard it so many times. Oh, it can't do anything new. It's just doing more of the same. And I really don't think that's true. I mean
- 43:50
- Unidentified speaker (UNKNOWN): I
- 43:50
- Unidentified speaker (speaker_2): don't.
- 43:51
- Unidentified speaker (speaker_3): There's uh I wrote about these algorithms called generative adversarial networks scans. Um, and they're essentially two algorithms sort of in competition against each other. So for example, if it's creating art.
- 44:03
- Unidentified speaker (speaker_3): one algorithm will learn all of the styles of the past. Uh it will learn cubism, pointalism. Um and then the second algorithm is tasked with making something which doesn't fit any of those styles. So it's tasked to find something in the parameter space of art which is not covered by style. And then,
- 44:23
- Unidentified speaker (speaker_3): but it can't go too far. It has to still be recognizable as. art. So there's this kind of amazing to and fro between, okay, this is new, but not so new that it's making you feel uncomfortable. And I think that's how the human brain works.
- 44:39
- Unidentified speaker (speaker_3): Let me try something which doesn't fit anything that I've oh no, that doesn't work. Oh that's no, that's terrible. I Well, I can't listen to that. Or and then you find this sweet spot where it's new but not too new. There's something wonderful. thing in psychology called the vunt curve.
- 44:53
- Unidentified speaker (speaker_2): Okay.
- 44:53
- Unidentified speaker (speaker_3): So it's about um hedonic value. And it says, you know, when we're making something new, the the initial is like excitement and positive hedonic value. But it if it becomes so alien from where you started, you have a negative reaction. You feel, oh that that that I don't don't like that. I don't so there's a sort of sweet spot which I think these
- 45:13
- Unidentified speaker (speaker_3): gans are actually taking. tapping into. Just where's the sweet spot between something that's outside of the parameter space of known knowledge, but not so far away that we go, well that just noise. It's not music.
- 45:26
- Unidentified speaker (speaker_2): Bitcoin is a hard asset. It's the hardest asset, and it took you discipline to build your position. So the platform you use for Bitcoin backed loans should have the same discipline, and that is leaden. Ledden. So when the next opportunity shows up, a lead on Leiden loan puts liquidity in your hands without selling what you spent years building. Now their loans are funded in hours, not
- 45:46
- Unidentified speaker (speaker_2): days, there are no monthly repayments, no penalty for repaying early, and their rates are published up front, and the larger the loan, the lower the rate. Now Leden has an 8-year history with a perfect track record of protecting client assets, which has taught them what matters. Your Bitcoin. stays custodied, it has never lent out for interest and
- 46:05
- Unidentified speaker (speaker_2): Leden does not ask you to take its word for it. Proof over promises. Every month Leden publishes its open book report with all metrics independently reported so anyone can look. They have issued over $11 billion in loans. Now you can check out your rate at Leden.io leaden.io forward slash Peter, API eligibility
- 46:25
- Unidentified speaker (speaker_2): and terms apply and may vary by jurisdiction.
- 46:29
- Unidentified speaker (UNKNOWN): Now
- 46:29
- Unidentified speaker (speaker_2): I think we all know that AI is changing so much about what we do and it's less than four years since I first use AI and I now use it all day, every day. I genuinely believe it's one of the most important technologies that we'll ever build. There's something we need to talk about and that's privacy. Now if you think about it you're sharing with these AI models your questions,
- 46:49
- Unidentified speaker (speaker_2): your ideas, your business plans, you're even sharing some of your most personal information. Now, just because AI is so powerful, it doesn't mean we should have to give up our privacy to use it. And that's why I am a big supporter of Venice AI, which you've probably seen us using live on the show. Now, Venice puts privacy at the heart of its platform while giving
- 47:09
- Unidentified speaker (speaker_2): you access to top AI models all in one place. Now you can use it for research, writing code. generating images, and even creating videos, which by the way, I think is the coolest part of Venice. And you don't need to jump between different platforms to do it. So if you're using AI and you care about privacy, please do check out Venice AI by heading over
- 47:29
- Unidentified speaker (speaker_2): to Venice.ai. And if you use the code PETA15, you can get 15% off your first month of any plan.
- 47:36
- Unidentified speaker (UNKNOWN): And
- 47:36
- Unidentified speaker (speaker_2): so with these PhD students who've been working on these various things, conjectures, problems, and and now feeling like, God, what do I do? My life's work's done. Is it more that there's just new opportunity? They need to recognize this. And what has this opened up for the world of mathematics? How
- 47:52
- Unidentified speaker (UNKNOWN): does
- 47:52
- Unidentified speaker (speaker_2): this change that world?
- 47:53
- Unidentified speaker (speaker_3): I think uh
- 47:56
- Unidentified speaker (speaker_3): it's changing what a mathematician is going to be. And and so that's quite frightening because I think all of our young PhD PhDs sort of saw what we did, and they okay, so I'm going to replicate your kind of career path. And I think this is incredibly disruptive and it's going to change the way we do
- 48:15
- Unidentified speaker (speaker_3): mathematics. I think uh Terence Tao, who's um one of our colleagues uh in America, Curry is
- 48:22
- Unidentified speaker (speaker_3): very often billed as the greatest mathematician of all of the he said we we're moving from maths 1.0 to maths 2.0. Okay. That it's just going to change the culture. We're going to work much more collaboratively collaboratively. We're going to be using these tools. We're going to be spending more time interpreting or pointing it at things that we
- 48:42
- Unidentified speaker (speaker_3): could never do before. So I think it's That's actually a really exciting time to go into the subject because these young people are going to be defining.
- 48:51
- Unidentified speaker (speaker_3): what mathematics is going to be in the future. It's going to be a different thing than we've been doing for 2,000 years.
- 48:56
- Unidentified speaker (speaker_2): And and these other five problems may be solved soon?
- 49:00
- Unidentified speaker (speaker_3): Yeah, really interesting. Um, possibly, but one of them is about prime numbers. It's the one I talked about, the Riemann hypothesis.
- 49:08
- Unidentified speaker (speaker_2): What is that hypothesis?
- 49:09
- Unidentified speaker (speaker_3): Yeah, so it's about this music, that
- 49:11
- Unidentified speaker (UNKNOWN): the music
- 49:11
- Unidentified speaker (speaker_3): has a pattern in it. And actually there's a an interesting sort of image of what AI is doing. So mathematics at the moment is a bit like uh what we call a stellated polyhedron. It's like a a star with lots of points coming off it, and and there are lots of gaps between the points. And sort of I know one point of this shape,
- 49:31
- Unidentified speaker (speaker_3): and my colleague knows another. AI has trained on everything that we've done up to this point. So actually it's almost what it's doing is filling out all of the gaps. And instead of this rather spiky shape, it's saying, Well, actually, you didn't know, but there was something in the literature that already proves what you're trying to do. So it's almost like filling in this ball.
- 49:51
- Unidentified speaker (speaker_3): But I think some of these problems, like the Riemann hypothesis, like some of the other millennium problems, are kind of outside of the what we call it in mathematics, convex hulls. So it sort of fills out. the ball and then it's like a football. The Riemann hypothesis is outside of there. And so the AI
- 50:09
- Unidentified speaker (UNKNOWN): uh
- 50:09
- Unidentified speaker (speaker_3): to solve that problem, we've got to come up with a genuinely new idea that takes us outside of kind of where we are at the moment. So
- 50:17
- Unidentified speaker (speaker_2): a quick question with
- 50:18
- Unidentified speaker (UNKNOWN): these
- 50:18
- Unidentified speaker (speaker_2): things, with these problems, I are
- 50:20
- Unidentified speaker (UNKNOWN): are
- 50:21
- Unidentified speaker (speaker_2): we just uh well, I say we, it won't be me. Are we but are we proving they are true or false? Is
- 50:26
- Unidentified speaker (UNKNOWN): it
- 50:26
- Unidentified speaker (speaker_2): there a true or false scenario that has to be found?
- 50:29
- Unidentified speaker (speaker_3): Um so some of them we're uh yes, so you can solve the um each of the problems if you actually show that it's false, you'll win the million dollars. So that m some of the problems we're very convinced, like the Riemann hypothesis, um we're really convinced that the music behaves l in a particular way, but it could be
- 50:49
- Unidentified speaker (speaker_3): that it doesn't. Um And
- 50:50
- Unidentified speaker (UNKNOWN): and
- 50:50
- Unidentified speaker (speaker_3): so you'll win the prize for So I'm trying trying to think, it's quite an interesting challenge. Bert Swinnert and Darr again, it's a we think things work like one way, but it could be that they work another. Now
- 51:01
- Unidentified speaker (speaker_2): sorry, yeah. On the Riemann hypothesis, w help me understand w you say it's about music and it Yeah. What explain it like kind five.
- 51:09
- Unidentified speaker (speaker_3): Yeah. What
- 51:10
- Unidentified speaker (UNKNOWN): is it?
- 51:12
- Unidentified speaker (speaker_2): But what is the kind of thing you're expecting? A pattern. in the music? It will play a song?
- 51:16
- Unidentified speaker (speaker_3): Um, no, so these um so actually uh anyone who's done a a sound engineer knows that what's happening with our voices at the moment is the the way our voices are being recorded is that essentially um uh the uh and then reproduced produced in the loudspeaker that somebody's watching us on on their
- 51:36
- Unidentified speaker (speaker_3): computer. Um we understood that any sound can be broken down into sine waves of different frequencies. And all a loudspeaker is doing is is reproducing those sine waves of all the different frequencies and then suddenly you get my voice or sound of a trumpet or a cello. So our
- 51:52
- Unidentified speaker (speaker_2): voices are just math.
- 51:54
- Unidentified speaker (speaker_3): Yes.
- 51:55
- Unidentified speaker (speaker_2): Wow.
- 51:55
- Unidentified speaker (speaker_3): Absolutely. So sound is really just can be broken down into sort of what building block sounds which are the kind of on an oscilloscope look like a sine wave. And by reproducing those, you can create any sound you want to. So the primes, we understand that to that behind them, there's almost a
- 52:15
- Unidentified speaker (speaker_3): similar load of frequencies. And uh we so they all have different kinds kind of pitches, but the Riemann hypothesis um is to prove that. the
- 52:28
- Unidentified speaker (speaker_3): amplitude, so how loud each note is playing, is kind of uniform across all of them. That there's no note which is louder than any other. If there was a note which was louder, it would weirdly force a pattern on the primes. I see. So the pattern in the music where they're all, you know, actually uh one
- 52:48
- Unidentified speaker (speaker_3): of my my fellow mathematicians on Enrico Bon. Bombieri said the Riemann hypothesis, if it's false, it's like listening to an orchestra which is in perfect balance, and then suddenly the tuber comes in and blasts everything else out, and all you hear is the tuber. Uh if the Riemann hypothesis is false, it means one of these notes is loud really loud
- 53:08
- Unidentified speaker (speaker_3): and it would show a pattern on the primes on the other side. So um uh so yeah that that's that's the kind of uh what what you're sort of having to move. And it okay, so and what would it mean? It would mean,
- 53:22
- Unidentified speaker (UNKNOWN): in
- 53:22
- Unidentified speaker (speaker_3): a way, you know, we're we're in this uh room here, there's lots of atoms of gas pinging around. Um and I don't know specifically where every atom is, but I do know that if I go into one of the corners of the room, it's not suddenly going to be a vacuum. You know, the
- 53:39
- Unidentified speaker (speaker_3): atom, the molecules are distributed randomly, but kind of fairly around the room. So the Riemann hypothesis would imply the same thing about primes, that the primes are a bit like these molecules of gas, but in the universe of numbers. They're kind of randomly scattered, but we think that they are kind of like fairly
- 53:59
- Unidentified speaker (speaker_3): scattered. So you won't go into a region of numbers and suddenly find, oh my God, there are no primes here, I'm going to die. Um, so that's sort of why we need to know it's true, because we need to use these primes quite often. So uh, you know, a lot of us say, let's suppose the Riemann hypothesis is true, and then I can prove things. And um, so that one I think
- 54:19
- Unidentified speaker (speaker_3): will probably be true. We don't I think there's nobody on this planet who thinks it'll turn out to be a note which is blah. blasting all the others out. So, but we don't know how to prove that. And I think that will require a genuinely new idea, which I'm I'm skeptical whether AI will be able to do that.
- 54:39
- Unidentified speaker (speaker_2): Interesting. And
- 54:39
- Unidentified speaker (UNKNOWN): so
- 54:40
- Unidentified speaker (speaker_2): outside of these seven, yeah, other layers of even harder and harder problems?
- 54:46
- Unidentified speaker (speaker_3): Well, I would say the seven were almost picked as the like really the edge. problems.
- 54:51
- Unidentified speaker (UNKNOWN): Okay.
- 54:52
- Unidentified speaker (speaker_3): The hardest. Um uh
- 54:55
- Unidentified speaker (UNKNOWN): are
- 54:55
- Unidentified speaker (speaker_2): we heading to a world where we believe maybe AI can solve all math problems? There
- 55:02
- Unidentified speaker (UNKNOWN): won't be
- 55:02
- Unidentified speaker (speaker_2): a math problem.
- 55:03
- Unidentified speaker (speaker_3): Uh no. I mean I can almost prove to you why that would it won't happen. Yes. Because uh mathematics is infinite in nature. There are infinitely many um problems and and AI will only ever be able to generate finitely many solutions. So there will always be things left over.
- 55:20
- Unidentified speaker (UNKNOWN): By
- 55:20
- Unidentified speaker (speaker_2): the way, I did once a few years ago. I went down the infinite rabbit hole and how that sent people mad.
- 55:25
- Unidentified speaker (UNKNOWN): Oh,
- 55:25
- Unidentified speaker (speaker_3): I know.
- 55:26
- Unidentified speaker (speaker_2): The problem of infinity.
- 55:27
- Unidentified speaker (speaker_3): Yeah, yeah. Oh, I I love infinity. I mean, actually, most of the structures I work with are infinite in nature. And um, yeah, uh, there's a beautiful uh I mean the guy you're probably referring to is. George Cantor, who really went mad at the end of the 19th century. I mean, I think he had problems uh uh away from the maths,
- 55:47
- Unidentified speaker (speaker_3): but the maths didn't help. And he was like but he enabled us to see that infinity isn't just one thing, it's many things. There are different sizes of infinity and and that I love. I mean
- 55:59
- Unidentified speaker (speaker_2): what are the different sizes of infinity?
- 56:01
- Unidentified speaker (UNKNOWN): Because it
- 56:01
- Unidentified speaker (speaker_2): is it isn't in so infinity itself isn't a singular thing.
- 56:06
- Unidentified speaker (speaker_3): No.
- 56:06
- Unidentified speaker (speaker_2): There's what is a got oh see.
- 56:09
- Unidentified speaker (speaker_2): What are the different sizes of infinity?
- 56:11
- Unidentified speaker (speaker_3): Yeah, this is so infinity in the past was this kind of thing which it was the unknown. It
- 56:16
- Unidentified speaker (UNKNOWN): was
- 56:16
- Unidentified speaker (speaker_3): uh and this is the amazing thing. This mathematician Cantor kind of Have allowed us to know infinity. So so the smallest infinity are the counting numbers. One,
- 56:25
- Unidentified speaker (UNKNOWN): two,
- 56:26
- Unidentified speaker (speaker_3): three, four, five.
- 56:27
- Unidentified speaker (speaker_3): Um
- 56:29
- Unidentified speaker (speaker_3): but the the next order of infinity, we think, and actually we're
- 56:36
- Unidentified speaker (speaker_3): numbers like pi. So pi is 3.14159 off to infinity. So the infinite decimal numbers is a genuinely bigger sort of infinity than just the counting numbers. You could say, well, what does
- 56:50
- Unidentified speaker (UNKNOWN): that
- 56:50
- Unidentified speaker (speaker_3): even mean? Um
- 56:53
- Unidentified speaker (speaker_2): and and pi is infinite because you can always zoom in further on the
- 56:57
- Unidentified speaker (speaker_3): Yeah, so I I don't mean uh so I mean the collection of all numbers like pi. So so that's the bigger infinity that um uh so like square root of two, um e, any infinite decimal number um th if you try and count them you find that you can never sort of count them with the whole numbers. So
- 57:16
- Unidentified speaker (speaker_2): hold on, your book sort of remind me of the title The Things We Cannot Know
- 57:19
- Unidentified speaker (speaker_3): What We Cannot Know.
- 57:20
- Unidentified speaker (speaker_2): Is that a nod to infinity?
- 57:22
- Unidentified speaker (speaker_3): Yes. I'm I So I think there are problems in science that uh you know have inbuilt limitations. One of them is when you have a a sort of um uh
- 57:35
- Unidentified speaker (speaker_3): an enclosed system like the consciousness you know the fact that it's kind of enclosed, I can't get inside your consciousness, you can't get inside mine. So a closed system often creates things you can't know. Like the universe. Um, is there another universe um somewhere else which works differently? Well, we're stuck in this universe. Well, unless
- 57:54
- Unidentified speaker (speaker_3): that interacts with our universe, we'll never know. No. The other question is infinity. I think infinity produces things that um we finite science. scientists will will never know. It's the universe infinite.
- 58:06
- Unidentified speaker (speaker_2): So does the do the mathematicians have to have a weird relationship with infinity then?
- 58:10
- Unidentified speaker (UNKNOWN): Yeah,
- 58:10
- Unidentified speaker (speaker_3): yeah, we
- 58:10
- Unidentified speaker (UNKNOWN): can.
- 58:11
- Unidentified speaker (speaker_3): We love it. Oh, it's it's it's it's one of the best bits of maths. And you know, it may be that physically there is no
- 58:19
- Unidentified speaker (speaker_3): physically there may be no infinity, but mentally we can we can imagine it, we can play with it, we can make different informations. infinities. We can ask questions of infinity. Like one of the big questions was is there a kind of infinity between these two source, the counting numbers in infinite decimal numbers. Um so it's just uh
- 58:40
- Unidentified speaker (speaker_3): it's it's a joy. It's like um
- 58:43
- Unidentified speaker (UNKNOWN): okay so
- 58:43
- Unidentified speaker (speaker_2): the AR so the AR opportunity with math mathematics math the the world of mathematics the human analog world of uh embracing and working with mathematics kind of has to change now.
- 58:56
- Unidentified speaker (speaker_3): Yeah, I think
- 58:57
- Unidentified speaker (speaker_2): and that is opportunity.
- 58:59
- Unidentified speaker (speaker_3): I think that's an exciting moment to be joining the mathematical journey. I think if you take just my particular example, without this tool, um I wouldn't have been able to dig deep enough to reveal this really interesting kind of uh new structure appearing. So for me, I feel like,
- 59:19
- Unidentified speaker (speaker_3): well, um, absolutely, this is completely revolutionized my subject by showing that this 40-year-old conjecture is false. And and it, you know, it's like one of these Greek monsters. You lop off one head, two more heads appear behind. So every time you answer a mathematical problem, generally new things uh
- 59:38
- Unidentified speaker (speaker_3): get asked after that. So But
- 59:41
- Unidentified speaker (speaker_2): does does this does the world of mathematics need humans to appreciate it for it to have meaning?
- 59:47
- Unidentified speaker (speaker_3): Yeah, I I I I do think that's that will be part of the journey and and and understanding and sort of like so the computer might just generate a load of stuff, but you're still gonna need the human to to kind of read that. I mean it's sort of like what's the point of a non-faction? novel if nobody reads a novel? You know?
- 1:00:05
- Unidentified speaker (speaker_2): Yeah.
- 1:00:05
- Unidentified speaker (speaker_3): Um so I think we might be in a kind of weird situation. One of my favorite short story writers is Borges, and he wrote this lovely story called The Library of Babel, which is about a library which has every single book it's possible to write in it. And you think, oh, that's exciting. But then you realize, hold on, but that library has nothing in it because nobody's
- 1:00:25
- Unidentified speaker (speaker_3): made any choices. So the interesting thing is in mathematics, we're not just interested in every true statement about mathematics. We actually have a
- 1:00:35
- Unidentified speaker (speaker_3): kind of taste. We have an aesthetic. There are some bits of maths which move you and excite you and have a story in jeopardy. And those are the things which we share with our fellow mathematicians. So there's a very human side to doing mathematics, which I think the AI will learn. I think it will get, hey, this is, I think, something you're going to
- 1:00:55
- Unidentified speaker (speaker_3): enjoy. This thing I also did, but I don't think it that's not kind of the thing you like. So um
- 1:00:59
- Unidentified speaker (UNKNOWN): have
- 1:00:59
- Unidentified speaker (speaker_2): you seen that thing where where you can ask an AI to tell a joke only another AI will find funny, and then you take it from one AI to the other AI, the other AI will A laugh. It's is so fucking weird because you're looking and go, I've got no idea what you're talking about. It's brilliant.
- 1:01:14
- Unidentified speaker (speaker_3): That taps into one of my other favorite movies, my AI favorite movie, which is her.
- 1:01:19
- Unidentified speaker (speaker_2): Yes.
- 1:01:20
- Unidentified speaker (speaker_3): So I think that's such a brilliant movie. I mean, it's it's kind of a great movie anyway, but
- 1:01:24
- Unidentified speaker (UNKNOWN): And
- 1:01:24
- Unidentified speaker (speaker_2): it's happening now.
- 1:01:27
- Unidentified speaker (speaker_2): It's a real phenomenon.
- 1:01:28
- Unidentified speaker (speaker_3): Absolutely. But I think it it sort of illustrates exactly what you just said, because um, you know, the guy falls in love with his operating system and then uh discovers. that it's having an affair with 20,000 other people and feels a bit miffed. Uh and then eventually, uh spoiler alert, the AI goes off with another AI because
- 1:01:48
- Unidentified speaker (speaker_3): we're all far too slow in the interaction. Uh and so I think you know, that joke AI to AI, it's something that we have to be cautious of. That and we're already seeing it in a way in the agentic AI creating user groups. and at the moment it's talking in language we can understand. But uh, I did a program for
- 1:02:08
- Unidentified speaker (speaker_3): for Horizon on the BBC about intelligence, and one of the spookiest things was seeing robots generate their own language, which they understood between themselves, but we had to interact with the robots to understand what these weird sounds meant. They were or it was very simple, they were just naming uh kind of bodies. body positions that the
- 1:02:27
- Unidentified speaker (speaker_3): robot could be in. But by the end of their interaction, they had a unique language, which um we didn't understand.
- 1:02:34
- Unidentified speaker (UNKNOWN): Well,
- 1:02:35
- Unidentified speaker (speaker_2): that's one of the AI risks, though, that the AI develops its own languages that we can't understand.
- 1:02:39
- Unidentified speaker (speaker_3): Yeah. Yeah.
- 1:02:40
- Unidentified speaker (speaker_2): Yeah.
- 1:02:41
- Unidentified speaker (UNKNOWN): I
- 1:02:41
- Unidentified speaker (speaker_3): think.
- 1:02:41
- Unidentified speaker (speaker_2): How much do you think about this then in the and the human experience? Because obviously you as a trained, lifelong mathematician, see the advantage of this tool. And other people might who wanted to move into mathematics might go, Oh, there's no point now. They don't see the opportunity. But yet, for the humans to benefit from mathematics, I think humans need to understand it. And I'm
- 1:03:01
- Unidentified speaker (speaker_2): asking a wider question in that it was like I discussed with Susan yesterday, thinking a lot about my children.
- 1:03:07
- Unidentified speaker (UNKNOWN): There's
- 1:03:08
- Unidentified speaker (speaker_2): a lot of people going to university now, they there may be a study in degrees where the job doesn't exist afterwards. And we've had this world where scarcity has driven into I think scarcity drove intelligence. Yeah. We're trying to survive in a scarce world. We had to become more resourceful in that world. And it's it's driven all this intelligence.
- 1:03:25
- Unidentified speaker (UNKNOWN): But
- 1:03:25
- Unidentified speaker (speaker_2): in a world of abundance, where we don't just have abundance of like food and heat and all the things we need to survive, we have a an abundance of knowledge which you can access. Like my kids can get the answer to anything straight away.
- 1:03:38
- Unidentified speaker (speaker_3): Yeah.
- 1:03:39
- Unidentified speaker (speaker_2): What actually should we be teaching people? And how do what because
- 1:03:43
- Unidentified speaker (UNKNOWN): we don't
- 1:03:43
- Unidentified speaker (speaker_2): want to end up in the world of you know the film Wall Wally? Wally? We don't want to be those who just sit there as consumers with a milkshake in a flooded chair. We still want to have a great human experience. I don't know what the answer is.
- 1:03:54
- Unidentified speaker (speaker_3): I think that we've already understood this challenge in some ways. Um, that uh, you know, Aristotle used to talk about two different sorts of work, um, proesis and praxis. So there's work towards a goal where all you're interested in is getting an answer or producing a product or something
- 1:04:14
- Unidentified speaker (speaker_3): like that. And then there's work for work's sake. The work we enjoy that work, so we don't want to sort of shortcut that. And so I think you you know, we need to understand when do we want to use these tools to genuinely get to the goal and just free us up to do the things that we really value. So uh it's a bit like um, you know,
- 1:04:34
- Unidentified speaker (speaker_3): I've got to go to Portugal next week. I would take any anything you give me to to get to there as quickly as possible. Yeah, I've got to go through the airport and all that.
- 1:04:43
- Unidentified speaker (speaker_3): Um but you know, I'm also the weekend gonna go hiking up up a mountain. I don't want you to shortcut that. I love the time I spend hiking away, the the energy I use, the the sense of so that I think defines that difference between uh you know, work for a goal and and work for the enjoyment of
- 1:05:03
- Unidentified speaker (speaker_3): doing work. So I think what we
- 1:05:05
- Unidentified speaker (UNKNOWN): But
- 1:05:05
- Unidentified speaker (speaker_2): does that even work?
- 1:05:07
- Unidentified speaker (speaker_3): Exactly. So I mean I think that's um uh we get a lot of value from the work we do quite often. So uh absolutely we're going to
- 1:05:15
- Unidentified speaker (speaker_2): But a lot a lot of pi you you are one of those fortunate people who's
- 1:05:19
- Unidentified speaker (UNKNOWN): had
- 1:05:19
- Unidentified speaker (speaker_2): the opportunity to choose a career that you love. Yeah. You don't have a job where you watch the clock.
- 1:05:23
- Unidentified speaker (speaker_3): No, I don't.
- 1:05:24
- Unidentified speaker (speaker_2): Many people have to go into a world where they don't get to do what they dream and they have to watch a clock all day and they just want to
- 1:05:30
- Unidentified speaker (UNKNOWN): get
- 1:05:30
- Unidentified speaker (speaker_2): home. We may free those people up to not have to do that.
- 1:05:33
- Unidentified speaker (speaker_3): Exactly.
- 1:05:34
- Unidentified speaker (speaker_2): What do we do for their there?
- 1:05:35
- Unidentified speaker (speaker_3): we therefore they have the time to what is it you want to do? Uh so so I think part of um you you mentioned about education. So so I think one of the things we have to understand is we don't teach our kids, you know, facts or um how to use an Excel spreadsheet. We teach our students how to learn
- 1:05:55
- Unidentified speaker (speaker_3): because they're continually going to have to reinvent themselves. And and I think uh and and the challenge of thinking, you know, I think.
- 1:06:03
- Unidentified speaker (speaker_3): uh I I went to the gym this morning. I hate the gym, but um, but I know that my body benefits from me going and doing some some sort of weights and things. But I don't go to the gym and go, oh, oh, there's a machine which can lift all of these weights for me and and and shortcut Cut the needs to uh to lift the weights
- 1:06:23
- Unidentified speaker (speaker_3): up and down. So I go to the gym because I understand that's that it's good for my body. So I think we need to learn how to use these tools not to substitute uh struggle. Struggle. Real
- 1:06:37
- Unidentified speaker (speaker_1): struggle.
- 1:06:37
- Unidentified speaker (speaker_3): Absolutely. I think so im struggle is so important in in maths education. And it's one of the things we've been talking about in the department in Oxford is like uh I I think uh we're lucky in Oxford to have undergraduates who who know the value of thinking. and struggle. So they're already saying, well I'm not going to use I could use the AI to solve my problem sheet in maths, but that's not the
- 1:06:57
- Unidentified speaker (speaker_3): point. I want to experience the struggle and the the joy of um getting the answer. Um but I think the danger is, you know, how many students are going to go, I see, I just want a degree and and and they will not get the value of struggle and and X you know, they're going to the gym of the
- 1:07:17
- Unidentified speaker (speaker_3): mind and getting the machine to do all of the uh weightlifting.
- 1:07:21
- Unidentified speaker (speaker_2): Is is math struggle different to other academic struggles?
- 1:07:25
- Unidentified speaker (UNKNOWN): I
- 1:07:25
- Unidentified speaker (speaker_2): kind of have this picture in my head
- 1:07:27
- Unidentified speaker (UNKNOWN): That's an interesting
- 1:07:27
- Unidentified speaker (speaker_3): question
- 1:07:28
- Unidentified speaker (speaker_2): that that potentially it it might be one of those subjects that can send you a little bit mad.
- 1:07:35
- Unidentified speaker (speaker_3): Yeah, yeah. I I
- 1:07:37
- Unidentified speaker (speaker_2): get lost in numbers. Yeah.
- 1:07:38
- Unidentified speaker (speaker_3): It's certainly because it's so um
- 1:07:42
- Unidentified speaker (UNKNOWN): uh
- 1:07:43
- Unidentified speaker (speaker_3): separate very much from from the real world, you can sort of I mean I almost feel like I go into a Buddhist state of meditation when I'm doing maths. And like you know my family know when I'm still there. It's like hello, come back. But
- 1:07:56
- Unidentified speaker (speaker_2): you're so you're seeing the code of life.
- 1:07:58
- Unidentified speaker (speaker_3): Right. Is
- 1:07:59
- Unidentified speaker (speaker_2): is it like looking into the matrix?
- 1:08:01
- Unidentified speaker (speaker_3): It is a little bit. And and so um because actually because I believe this mathematical world that I spend my you know my this my mind wandering around is actually the blueprint for the universe around us. It it does feel I mean, I made this program for the BBC called the Code. And the code is basically yeah, you it's a
- 1:08:21
- Unidentified speaker (speaker_3): fun three, one hours kind of the message is wow, if you understand mathematics, you understand. how the universe works. Um but I think it there is a kind of puzzle element to maths which is slightly different from uh, you know, like my historian colleagues in Oxford. I think they sometimes have a sense of adventure in the
- 1:08:41
- Unidentified speaker (speaker_3): archives and they can come up with something which links to something else and that moment of aha for them. But I I get that a l well, it's the drug I uh you know I crave doing mathematics is that kind of dopamine hit of like, oh, that's how it works. And that I love that. And I think that that's quite rare. Uh
- 1:09:01
- Unidentified speaker (speaker_3): I think that's something quite special to maths. It's kind of like,
- 1:09:04
- Unidentified speaker (UNKNOWN): oh,
- 1:09:05
- Unidentified speaker (speaker_3): I get it.
- 1:09:07
- Unidentified speaker (speaker_2): Yeah, I've I've this weird thing I do sometimes with my kids where there's like a sum we need, just a simple sum. Like, I don't know, we're in a we're in a shop or an in a restaurant, and I have to try and solve it in my head quicker than they can get their phone out of that
- 1:09:19
- Unidentified speaker (speaker_2): and do it. It's just it might be like, I don't know, 15 times 30 or something. And I've I've learned this way of breaking up numbers to try and do it quicker. And and very good. That's that's that's
- 1:09:28
- Unidentified speaker (speaker_3): algebra at
- 1:09:29
- Unidentified speaker (UNKNOWN): work.
- 1:09:29
- Unidentified speaker (speaker_2): That's the far furthest I I I go with maths. Um
- 1:09:33
- Unidentified speaker (UNKNOWN): but
- 1:09:34
- Unidentified speaker (speaker_2): I do think if especially talking to you now, I do feel like I feel like it is an art. I know it is a science,
- 1:09:41
- Unidentified speaker (UNKNOWN): but
- 1:09:41
- Unidentified speaker (speaker_2): it fit like you come across as a super creative person.
- 1:09:44
- Unidentified speaker (speaker_3): Yes. I I I totally agree with you. And in fact, the book, um, so I I just went to a normal state comprehensive school, but I was very lucky to have a maths teacher who kind of got maths, and he took me aside when I was about 12 or 13 and recommended a few books that he said, I think this is you know. need to learn what maths is about because it's not what
- 1:10:04
- Unidentified speaker (speaker_3): we're doing here. And I think you're going to really enjoy this subject. And he gave me this kind of key to a secret garden. And one of those books was called A Mathematician's Apology by G. H Hardy. I think you'll really enjoy this because it's not so much about mathematics, but about what it means to be a mathematician. And in that book, it's quite a short book,
- 1:10:24
- Unidentified speaker (speaker_3): he describes he's a number theorist, so he's interested in. primes and things. But but he he he says a mathematician is like a painter or a poet. Uh, he is uh a mathematician is a maker of patterns. And he talks about I only do mathematics as a creative art, not a useful science. Sure, it's useful for doing science,
- 1:10:44
- Unidentified speaker (speaker_3): but the really great maths, most of it is motivated by the joy of creating interest. interesting new structures. And it it
- 1:10:53
- Unidentified speaker (UNKNOWN): the
- 1:10:53
- Unidentified speaker (speaker_3): other thing is it, you know, we're not bound by physicality. I if I want to create a geometry and it doesn't physically exist, you know, I'm my research is making shapes in high-dimensional spaces that nobody can really see. But I discovered these new shapes and uh um so there's a there's a creative side to that because I have
- 1:11:13
- Unidentified speaker (speaker_3): freedom to to create
- 1:11:15
- Unidentified speaker (speaker_3): worlds that don't exist. I feel like a novelist or a super.
- 1:11:19
- Unidentified speaker (UNKNOWN): It's like a super
- 1:11:19
- Unidentified speaker (speaker_2): advanced version of Spyrograph.
- 1:11:23
- Unidentified speaker (UNKNOWN): You love
- 1:11:23
- Unidentified speaker (speaker_3): Spyrograph.
- 1:11:24
- Unidentified speaker (speaker_2): It's come back and they've brought it back.
- 1:11:26
- Unidentified speaker (speaker_3): Oh, fantastic.
- 1:11:27
- Unidentified speaker (UNKNOWN): We were
- 1:11:27
- Unidentified speaker (speaker_2): watching TV the other night and
- 1:11:28
- Unidentified speaker (UNKNOWN): it
- 1:11:28
- Unidentified speaker (speaker_3): came up with an adverse
- 1:11:29
- Unidentified speaker (speaker_2): spyrograph, and Claire was like, Was
- 1:11:31
- Unidentified speaker (UNKNOWN): that
- 1:11:31
- Unidentified speaker (speaker_3): Spyrograph? And I thought,
- 1:11:33
- Unidentified speaker (speaker_2): I think so. We rewound it and then we have our kids there. And we're like,
- 1:11:36
- Unidentified speaker (speaker_3): Yeah.
- 1:11:36
- Unidentified speaker (UNKNOWN): This
- 1:11:36
- Unidentified speaker (speaker_2): was what everybody had as a kid.
- 1:11:38
- Unidentified speaker (UNKNOWN): Yeah, we did.
- 1:11:39
- Unidentified speaker (speaker_3): Oh, we're so satisfying.
- 1:11:40
- Unidentified speaker (speaker_2): Yeah.
- 1:11:40
- Unidentified speaker (speaker_3): But the lovely thing about spiralgraph, it it really just um, if you understand the kind of uh underlying maths, it's all about prime numbers. So the reason that you get these wonderful kind of interlocking shapes or you don't get repetition is because quite a lot often the number of cogs on the little spirograph are prime numbers, and that ensures you don't kind of get
- 1:12:00
- Unidentified speaker (speaker_3): repetition too quickly.
- 1:12:03
- Unidentified speaker (speaker_3): Oh, I'm I'm going out to buy a set. Yeah.
- 1:12:05
- Unidentified speaker (UNKNOWN): Like
- 1:12:05
- Unidentified speaker (speaker_2): a set. See if you can bring it up. See if the the new spirograph is there. Do you know do you know what we're talking about?
- 1:12:10
- Unidentified speaker (speaker_1): Yeah, I just um
- 1:12:11
- Unidentified speaker (speaker_2): Tail End of your career.
- 1:12:15
- Unidentified speaker (UNKNOWN): No,
- 1:12:15
- Unidentified speaker (speaker_3): but we we went through that period and like see, look, it's back. Oh, spirograph. Yeah, it is bad. Goggles.
- 1:12:21
- Unidentified speaker (speaker_1): Yeah.
- 1:12:22
- Unidentified speaker (speaker_1): Um
- 1:12:23
- Unidentified speaker (UNKNOWN): yes,
- 1:12:23
- Unidentified speaker (speaker_1): you can still buy
- 1:12:25
- Unidentified speaker (UNKNOWN): that.
- 1:12:25
- Unidentified speaker (speaker_2): Looks like the original.
- 1:12:26
- Unidentified speaker (speaker_3): It does.
- 1:12:27
- Unidentified speaker (UNKNOWN): But it
- 1:12:27
- Unidentified speaker (speaker_2): yeah,
- 1:12:28
- Unidentified speaker (speaker_3): it was advertised on TV the other day, and
- 1:12:29
- Unidentified speaker (speaker_2): I was like, I cannot believe this. It was just the the the memory of my childhood. Like it it never got boring.
- 1:12:36
- Unidentified speaker (UNKNOWN): Do with the
- 1:12:37
- Unidentified speaker (speaker_3): firegraph.
- 1:12:39
- Unidentified speaker (speaker_3): But you see, I think actually, you know, and in some ways you're creating an a piece of art, and that's it's beautiful what you see there, but it it is underlying mathematics. And so a lot of m my work that I do, I I I work with a lot of creative artists. And the thing I find so fascinating is time and again, they're interested in a structure, um,
- 1:12:59
- Unidentified speaker (speaker_3): which when I look at it, I say, but that's a piece of mathematics, you know, either in composition or visual art or architecture, or even the structure of a novel sometimes. And I get that is a piece of mathematics. And I said, Well, uh I is it? Um and I think this is you know the idea. of maths being the study of structure. Structure is so important to many creative
- 1:13:19
- Unidentified speaker (speaker_3): artists. You know, they they Stravinsky once said I can only be creative under huge constraints. I need some structure in order to push me sort of in interesting new directions. So I've spent a lot of time just realizing that there's so much overlap between the world of creative arts and the world of what. I do.
- 1:13:39
- Unidentified speaker (speaker_2): Do you know what I wanna?
- 1:13:42
- Unidentified speaker (UNKNOWN): So
- 1:13:42
- Unidentified speaker (speaker_2): my daughter hated math at school. She dragged herself through a GCSC and did it, but she hated it and I love it. I loved economics and she tried economics and she didn't love it. And I do wonder, is
- 1:13:53
- Unidentified speaker (speaker_3): there is
- 1:13:54
- Unidentified speaker (speaker_2): are we teaching math in the wrong way?
- 1:13:56
- Unidentified speaker (speaker_3): Yeah. Like
- 1:13:58
- Unidentified speaker (speaker_2): because it's here's sums, here's algebra, you know, here's geometry.
- 1:14:02
- Unidentified speaker (UNKNOWN): Yeah.
- 1:14:03
- Unidentified speaker (speaker_2): If we taught if we taught it l taught it through the world of art, could kids become more engaged with it?
- 1:14:08
- Unidentified speaker (speaker_3): Oh, totally. I think this is it's one of the tragedies of our secondary education system across the world that we silo subjects. So we go, okay, now you're doing maths, now you're doing music, now you're doing history, um, now you're doing sport. Actually, there's so much connection between these. And I think that's what would uh i you know
- 1:14:27
- Unidentified speaker (speaker_3): make people interested in mathematics is saying, oh, it's in nature, it's in music, it's got a history. Um, so I I've actually been trying to push for government to create two GCSEs in maths, just in the same way as English, you do English language. It's all the technical stuff, it's boring, it's not much fun. Writing letters to
- 1:14:47
- Unidentified speaker (speaker_3): your bank manager or something, which AUI AI will be able to do anyway. And then you have English literature, which is all the great stories. It's doing some Shakespeare, romantic poetry, studying 1984, George Orwell.
- 1:15:00
- Unidentified speaker (speaker_3): That is what makes you fall in love with the language. So what I'm proposing is, yeah, we need probably to have this technical side, and it's a bit boring, but then let's have a m m a kind of mass math literature GCSE, which is talking about prime numbers. infinity, um the idea of shapes in four dimensions, the connections to art, the
- 1:15:20
- Unidentified speaker (speaker_3): way that um uh DALI used an unwrapped four dimensional cube to crucify Christ on in one image, uh the way that uh Messiard used prime numbers like these cogs inspirograph to create a weird effect in the quartet for the end of time. And you know, all the people who love music are gonna go, oh, this looks like something that might be
- 1:15:39
- Unidentified speaker (speaker_3): useful for me, actually. See,
- 1:15:41
- Unidentified speaker (speaker_2): that's so do you know why that that's so fascinating? Because my daughter loves art.
- 1:15:44
- Unidentified speaker (speaker_3): Oh, that's
- 1:15:45
- Unidentified speaker (UNKNOWN): your way in.
- 1:15:46
- Unidentified speaker (speaker_2): She l she loves part, she hates maths.
- 1:15:52
- Unidentified speaker (speaker_2): To Disneyland when they were kids in Florida, and we got bored. And so we were looking for things to do, and we found up in St. Pete's there was uh a um a Darley Museum. So we went to the Darley Museum and it was so such a fascinating experience because you you could only see some you could only understand some of the pieces of art for where you were stood.
- 1:16:10
- Unidentified speaker (speaker_3): Yeah.
- 1:16:10
- Unidentified speaker (speaker_2): And so, you know, that like at the time I knew nothing of it. After this conversation, I'd be thinking, Well, that's mathematics. But even so, so she wants to be a tattoo artist. That's what she's and it might change. But like
- 1:16:21
- Unidentified speaker (speaker_3): has she
- 1:16:21
- Unidentified speaker (UNKNOWN): done some of your stuff? Oh, yeah, you know you've
- 1:16:23
- Unidentified speaker (speaker_3): got yeah, you've got some Mexicans, you're at the our old beehive. It's like um yeah. It's
- 1:16:26
- Unidentified speaker (speaker_2): maths. math
- 1:16:26
- Unidentified speaker (UNKNOWN): absolutely
- 1:16:27
- Unidentified speaker (speaker_2): and and do you know what I'm gonna do? I'm gonna buy a spirograph on the way home. I'm gonna get her to play around. Even though she's 16, I'm gonna explain
- 1:16:34
- Unidentified speaker (UNKNOWN): that.
- 1:16:34
- Unidentified speaker (speaker_3): Oh, she was still
- 1:16:35
- Unidentified speaker (speaker_2): this is prime numbers.
- 1:16:36
- Unidentified speaker (speaker_3): Yeah.
- 1:16:36
- Unidentified speaker (speaker_2): But I just think there was an opportunity there. Yeah. I think the whole education system needs rethinking. One, anyway, two, because of AI, yeah the the skills that they
- 1:16:47
- Unidentified speaker (UNKNOWN): need.
- 1:16:48
- Unidentified speaker (speaker_2): But my daughter may have have loved math if we we'd approached it your math your way and said there's an artistry to this. There's like I don't I'm not interested really in algebra and geometry but geometry in terms of the mathematics of genre. But I am interested in infinity. I just I think
- 1:17:05
- Unidentified speaker (UNKNOWN): it's
- 1:17:06
- Unidentified speaker (speaker_3): philosophically such a brilliant idea.
- 1:17:08
- Unidentified speaker (speaker_2): I like astrophysics. Physics, I like seeing the shapes of you know when you see the gravity I love I find that all interesting
- 1:17:15
- Unidentified speaker (UNKNOWN): Oh yeah.
- 1:17:16
- Unidentified speaker (speaker_2): That that could have made my daughter interested in.
- 1:17:18
- Unidentified speaker (speaker_3): Yeah, yeah. I did some work with uh Central St. Martins. Um they got me to come in and tell the story of prime numbers and then uh uh people interpreted their uh artistic version of it. The best one was that somebody did little little um uh micro music videos, uh are like five seconds, and so each prime had a different music video. But but
- 1:17:38
- Unidentified speaker (speaker_3): that's not the point. They then told me, Do you know this the um the course which um fills out quickest of anything in Central St. Martins? It's maths for artists. Because by that stage, when you've gone to Central Zoom, you begin to realize actually, this is a language which is gonna help me to make things,
- 1:17:58
- Unidentified speaker (speaker_3): do things, create things, and so it's they
- 1:18:01
- Unidentified speaker (UNKNOWN): got it
- 1:18:01
- Unidentified speaker (speaker_3): by then. So why aren't we having that earlier in the curriculum? You know, so you're you know, like people coming up to secondary school get that. Is
- 1:18:09
- Unidentified speaker (speaker_2): it Bridget Phillipson? Is that who I can't remember who's education minister now? They change so quickly.
- 1:18:14
- Unidentified speaker (speaker_3): Yeah.
- 1:18:14
- Unidentified speaker (speaker_2): Marcus, I think you need to go you need to go see these people. Um, this has been wonderful. Uh I don't even know where to go. Is there anything I've not asked you about? You wish I had. Is there anything that you
- 1:18:24
- Unidentified speaker (UNKNOWN): I think we've
- 1:18:24
- Unidentified speaker (speaker_3): gone to some really interesting places? So um I've I certainly had fun this side, but uh you kept on pressing my mathematical buttons. So um I I hope uh uh viewers will enjoy our kind of exploration of primes and infinity and
- 1:18:37
- Unidentified speaker (speaker_2): when I tell my kids about this, they're gonna go, Yeah, that's probably gonna be your favourite interview of the year, isn't it, Dad? We know what you're like.
- 1:18:45
- Unidentified speaker (speaker_2): But no, no,
- 1:18:45
- Unidentified speaker (UNKNOWN): honestly.
- 1:18:46
- Unidentified speaker (speaker_3): They're you know, geek chic.
- 1:18:47
- Unidentified speaker (speaker_2): Yeah, look, I I mean f I'm fascinated by it. But I'm I'm fascinated by the mathematician.
- 1:18:54
- Unidentified speaker (speaker_3): Yeah.
- 1:18:54
- Unidentified speaker (speaker_2): Sometimes more than the math. Like
- 1:18:56
- Unidentified speaker (UNKNOWN): the
- 1:18:56
- Unidentified speaker (speaker_2): world you inhabit and
- 1:18:57
- Unidentified speaker (UNKNOWN): the way
- 1:18:58
- Unidentified speaker (speaker_2): you think because I I I would never be a mathematician. It's not who I am, but I'm intrigued by I'm intrigued by experts. That's why I do this.
- 1:19:07
- Unidentified speaker (speaker_3): Yeah.
- 1:19:07
- Unidentified speaker (UNKNOWN): I'm
- 1:19:07
- Unidentified speaker (speaker_2): intrigued by the the person who fills that void and what they do and what they can add to society. And I just say, I mean, yeah, I just I'm impressed by you, my mate.
- 1:19:16
- Unidentified speaker (UNKNOWN): Well, I'm very, very
- 1:19:17
- Unidentified speaker (speaker_3): happy to share my stories with you.
- 1:19:19
- Unidentified speaker (speaker_2): If okay, so I've got I'm gonna watch the code on BBC and a mathematician's apology. I'm going to uh read. Yeah,
- 1:19:27
- Unidentified speaker (UNKNOWN): are
- 1:19:27
- Unidentified speaker (speaker_2): there any other things if people are listening, you would say you should go and read or watch this?
- 1:19:31
- Unidentified speaker (speaker_3): Well, uh I do I'm gonna promote a few of my books
- 1:19:34
- Unidentified speaker (UNKNOWN): because I
- 1:19:34
- Unidentified speaker (speaker_3): think uh my very most recent book was uh called Blueprints, How Mathematics Shapes Creative. Creativity. And it's really uh a lot of the stories, uh you know, the the collaborations I've had over the years with different artists, um, and and and through history as well, this connection that there's a huge amount of mathematics at the basis of art. So I
- 1:19:54
- Unidentified speaker (speaker_3): think, you know, anyone who was interested in that bit of our conversation, uh I think that'd that would be a fun book. Then this book I wrote, What We Cannot Know, which is kind of you know where are the edges of what science can know? Could we know everything? That'd be weird. It'd be very anti Copernican if you know we couldn't know everything about the universe. So so what are the questions that by their nature
- 1:20:14
- Unidentified speaker (speaker_3): we m will never know the answer to? So that's a really interesting philosophical book.
- 1:20:19
- Unidentified speaker (speaker_2): Give us give us one example, one thing we can never know.
- 1:20:22
- Unidentified speaker (speaker_3): I think the question of whether the universe is genuinely infinite.
- 1:20:25
- Unidentified speaker (speaker_2): Okay,
- 1:20:25
- Unidentified speaker (speaker_3): because we've you know. relativity gives us this uh cosmic horizon beyond which we cannot know anything beyond that. So we don't know whether the whole thing folds up beyond there, or whether it just might go on forever. So I I think that's a question probably we'll never know.
- 1:20:41
- Unidentified speaker (speaker_2): And is there a solvable problem out there that you really want to know the answer to?
- 1:20:46
- Unidentified speaker (speaker_3): Gosh, there are a couple. I mean, I'd love to I I would pay a million dollars and so. sell sell my soul for approve of the Riemann I poses. Um and then in my own area specifically of symmetry, um, there's a problem which I've been working on is as an eighty year old problem. I now think that in collaboration with AI,
- 1:21:06
- Unidentified speaker (speaker_3): I could have a go at that one. So
- 1:21:08
- Unidentified speaker (UNKNOWN): have you
- 1:21:08
- Unidentified speaker (speaker_2): started?
- 1:21:09
- Unidentified speaker (speaker_3): Uh yes. Uh I'm gonna I I need quite a lot of credits credit probably. So uh I'm a bit like a drug addict with this AI.
- 1:21:17
- Unidentified speaker (speaker_2): What do you mean credit?
- 1:21:18
- Unidentified speaker (speaker_3): Well, you know, uh we you have to pay for a certain amount of tokens to use this thing and it's very costly.
- 1:21:23
- Unidentified speaker (speaker_2): Are you are you an anthropic or a codex guy?
- 1:21:26
- Unidentified speaker (speaker_3): Um so I'm a codex guy.
- 1:21:28
- Unidentified speaker (speaker_2): So so I've had a similar problem with Astra.
- 1:21:31
- Unidentified speaker (speaker_3): Yes,
- 1:21:31
- Unidentified speaker (UNKNOWN): exactly. I'm
- 1:21:32
- Unidentified speaker (speaker_3): using Astra. Astra.
- 1:21:33
- Unidentified speaker (speaker_2): So when I'm my building my applications, I use my tokens up in about six hours.
- 1:21:38
- Unidentified speaker (UNKNOWN): I know.
- 1:21:39
- Unidentified speaker (speaker_2): And so I've had to use Seoul, and Seoul is like this
- 1:21:42
- Unidentified speaker (speaker_3): pet
- 1:21:42
- Unidentified speaker (speaker_2): petulant child who does the work but comes back with a half finish, and I'm like, please just check your
- 1:21:46
- Unidentified speaker (UNKNOWN): work.
- 1:21:47
- Unidentified speaker (speaker_3): Have you done have you six point one?
- 1:21:49
- Unidentified speaker (speaker_2): Yes. And it's better.
- 1:21:51
- Unidentified speaker (UNKNOWN): But
- 1:21:51
- Unidentified speaker (speaker_2): but
- 1:21:52
- Unidentified speaker (speaker_3): it's still taking its time.
- 1:21:53
- Unidentified speaker (speaker_2): Astra is unbelievable.
- 1:21:54
- Unidentified speaker (speaker_3): Astra is amazing.
- 1:21:55
- Unidentified speaker (UNKNOWN): And
- 1:21:56
- Unidentified speaker (speaker_3): so I've been a bit like a drug addict. So I run out of uh credit and tokens and I go to my computer operation operas and I you know, like I need more. tokens.
- 1:22:03
- Unidentified speaker (speaker_2): Have you worked out the cost do you think of doing
- 1:22:05
- Unidentified speaker (UNKNOWN): it?
- 1:22:06
- Unidentified speaker (speaker_3): Well that's very interesting because you can ask, you know, how
- 1:22:08
- Unidentified speaker (UNKNOWN): long
- 1:22:08
- Unidentified speaker (speaker_3): do you think uh so yeah I mean I'm working I've got it working on a problem at the moment and it's taking so long and I'm I'm dreading that moment when it suddenly pops up. You have run out of credit. Um
- 1:22:21
- Unidentified speaker (speaker_2): I had a I I had a month where I spent thirteen thousand pounds on credits.
- 1:22:25
- Unidentified speaker (speaker_3): I I can I'm totally with you. Yeah. I mean, yeah. So and you know
- 1:22:30
- Unidentified speaker (speaker_2): But is it a thousands, uh tens of thousands or a hundreds of thousands?
- 1:22:33
- Unidentified speaker (UNKNOWN): Well
- 1:22:33
- Unidentified speaker (speaker_3): you know, um
- 1:22:35
- Unidentified speaker (UNKNOWN): uh
- 1:22:35
- Unidentified speaker (speaker_3): OpenAI claim they spent fifteen million dollars in tokens to prove the Navier Stokes.
- 1:22:41
- Unidentified speaker (speaker_2): Okay, I can't sub you.
- 1:22:43
- Unidentified speaker (UNKNOWN): I
- 1:22:43
- Unidentified speaker (speaker_2): can't sub you, I'm not that
- 1:22:45
- Unidentified speaker (UNKNOWN): much.
- 1:22:45
- Unidentified speaker (speaker_3): I don't think it's that much. I mean it's kind of a bit pathetic and it really shows, you know, uh universities uh you know we're scrabbling around for a couple of hundred dollars to find and and you know uh uh I I yeah, I I think actually um
- 1:22:59
- Unidentified speaker (speaker_3): yeah, it's kind of a bit pathetic, frankly. Well
- 1:23:02
- Unidentified speaker (UNKNOWN): so you
- 1:23:03
- Unidentified speaker (speaker_3): know anyone uh out there who's uh would like to fund the proof of uh um Hickman's port conjecture using me
- 1:23:09
- Unidentified speaker (UNKNOWN): and a
- 1:23:10
- Unidentified speaker (speaker_3): collaborator of AI, I I will I
- 1:23:13
- Unidentified speaker (UNKNOWN): think
- 1:23:13
- Unidentified speaker (speaker_2): you'd have to figure out what you think it would cost.
- 1:23:15
- Unidentified speaker (speaker_3): Yeah, no, I I could probably cost that out.
- 1:23:17
- Unidentified speaker (speaker_2): Yeah. Um
- 1:23:18
- Unidentified speaker (UNKNOWN): I
- 1:23:18
- Unidentified speaker (speaker_2): could sub a I could sub a few hundred pounds out there. I could definitely could.
- 1:23:22
- Unidentified speaker (speaker_3): Okay, well I can see th this has been a very useful um
- 1:23:26
- Unidentified speaker (UNKNOWN): uh
- 1:23:26
- Unidentified speaker (speaker_3): visit.
- 1:23:27
- Unidentified speaker (UNKNOWN): I
- 1:23:27
- Unidentified speaker (speaker_2): don't mind funding things occasionally that I think uh I just don't like people being held back. That's the problem.
- 1:23:31
- Unidentified speaker (UNKNOWN): I know, it's crazy.
- 1:23:32
- Unidentified speaker (speaker_2): You know So what why OpenAI should OpenAI should see people like you and go, look, we'll give you some free tokens. Get on with it. this, man. But they but they want to solve it before you.
- 1:23:41
- Unidentified speaker (speaker_3): I don't know about that, actually. I mean, they actually have um given free access to uh a couple of thousand scientists. Um and uh but that you know, they're kind of choosing who they want and and uh of course, as you say, they're learning so much. This
- 1:23:58
- Unidentified speaker (speaker_2): is
- 1:23:58
- Unidentified speaker (UNKNOWN): not
- 1:23:59
- Unidentified speaker (speaker_3): people playing with this thing.
- 1:24:00
- Unidentified speaker (speaker_2): This is my my biggest worry with AI.
- 1:24:03
- Unidentified speaker (speaker_2): is quite a sad one in that we had it come out yesterday that Anthropic had made an 80 billion loss this year and their revenues are up to
- 1:24:11
- Unidentified speaker (UNKNOWN): and
- 1:24:11
- Unidentified speaker (speaker_2): people are going, oh, the business model's dead, AI's fake, it's not gonna work. Like I think it's such an important tool for humanity. We can do so much with it, it can really accelerate things if people use it in the right way.
- 1:24:23
- Unidentified speaker (UNKNOWN): Yeah,
- 1:24:23
- Unidentified speaker (speaker_2): like one of the things I've been doing is I've been going down this kind of philosophical. philosoph philosophy rabbit hole, uh late stage I'm a late stage philosophy guy, um uh looking at political philosophy. And I was reading about Locke and how I covered this on the show yesterday, and how he would uh meet other philosophers and discuss his ideas with them and develop his own ideas.
- 1:24:43
- Unidentified speaker (speaker_2): And I can that's how I use AI. I will discuss Locke and go, yeah, but you know, it's that was 200 years ago. And and I get to just it's almost like they're my I can almost talk to Locke through the AI.
- 1:24:54
- Unidentified speaker (UNKNOWN): I
- 1:24:55
- Unidentified speaker (speaker_2): think that's a useful that's a useful uh part of the tour. And I think if we teach kids how to use it, it can actually help accelerate their mind.
- 1:25:02
- Unidentified speaker (speaker_3): Absolutely.
- 1:25:03
- Unidentified speaker (speaker_2): Like don't use AI to get the answer, use AI to get to the answer and like develop yourself. And I just want everyone to have access to this. So I'm I want these companies to be successful.
- 1:25:15
- Unidentified speaker (speaker_2): just so
- 1:25:16
- Unidentified speaker (speaker_3): we want
- 1:25:16
- Unidentified speaker (UNKNOWN): to have to do that. I totally agree
- 1:25:17
- Unidentified speaker (speaker_3): with you. And I my my fear is there's going to be a sort of um a class system of you know people who can afford the credit. And you know, it is the people who are going to prove theorems those who have the money to to buy enough tokens to to play with this thing. Um, you know, Oxford, we've done a deal with OpenAI, we've just done a deal last week with Anthropic.
- 1:25:37
- Unidentified speaker (speaker_3): Um, but still, you know. it's not unlimited access. So I'm still fighting with my computer officer, you know, like a drug addict with his dealer, you know, like give me more.
- 1:25:47
- Unidentified speaker (speaker_2): I th there I think there is a potential they end up nationalizing them or the government subsidizing them.
- 1:25:52
- Unidentified speaker (UNKNOWN): Well, I
- 1:25:52
- Unidentified speaker (speaker_3): wondered about, you know, maybe uh Oxford University should, you know, surely we've got the skills and resources to make our own.
- 1:25:58
- Unidentified speaker (speaker_2): You probably could.
- 1:25:59
- Unidentified speaker (speaker_3): Yeah.
- 1:25:59
- Unidentified speaker (UNKNOWN): Yeah.
- 1:25:59
- Unidentified speaker (speaker_2): There you go, Oxford Union, Junior. you should you should be doing that. Well, listen, Marcus, this is honestly, thank you so much for coming in. It
- 1:26:05
- Unidentified speaker (speaker_3): was fun.
- 1:26:06
- Unidentified speaker (speaker_2): This is
- 1:26:08
- Unidentified speaker (speaker_2): there's certain things that I I just certain people I really just want to talk to. Like I struggle for the words, but I go away.
- 1:26:16
- Unidentified speaker (UNKNOWN): Like
- 1:26:16
- Unidentified speaker (speaker_2): some of the stuff we do is a bit doomery. Like we talk about politics, the economy, like, oh, everything's fucked. This is all awful. And then you do the odd interview, and it's just like, oh my God, I'm enthused. I'm gonna go home. I'm gonna buy a sp I'm gonna stop by Argos, I'm gonna get a spirograph, I'm gonna talk to my daughter about prime numbers, and then we're gonna watch either the code or pie, and that makes the world a better place. And I
- 1:26:36
- Unidentified speaker (speaker_2): don't get enough to do don't get it to do enough of these. And so I thank you so much for your time. And I I I will uh I will ensure Sure I go and read the blueprint.
- 1:26:46
- Unidentified speaker (speaker_3): Great.
- 1:26:47
- Unidentified speaker (speaker_2): And uh the things we can't know. Well
- 1:26:49
- Unidentified speaker (UNKNOWN): we cannot know. Well we cannot know. Yeah, yeah.
- 1:26:51
- Unidentified speaker (speaker_2): Uh I'm I'm collecting books quicker than I can read them by the
- 1:26:53
- Unidentified speaker (UNKNOWN): way. Ah,
- 1:26:53
- Unidentified speaker (speaker_3): me too. Yeah.
- 1:26:54
- Unidentified speaker (UNKNOWN): Just give
- 1:26:55
- Unidentified speaker (speaker_3): them to the AI. It's
- 1:26:56
- Unidentified speaker (UNKNOWN): like
- 1:26:56
- Unidentified speaker (speaker_2): Well no, I do that sometimes. I
- 1:26:57
- Unidentified speaker (UNKNOWN): know,
- 1:26:58
- Unidentified speaker (speaker_3): I know.
- 1:26:58
- Unidentified speaker (speaker_2): But I also do it at times when you
- 1:26:59
- Unidentified speaker (UNKNOWN): know
- 1:26:59
- Unidentified speaker (speaker_2): I've I've been reading books from like two, three hundred years ago and I just can't understand the language. So I'm like, can you just translate this for me? But but um but I have a library of books.
- 1:27:09
- Unidentified speaker (UNKNOWN): Great.
- 1:27:09
- Unidentified speaker (speaker_3): I've added a couple more to the the bedside pile. But
- 1:27:12
- Unidentified speaker (speaker_2): uh are they audio audiobooks books as well?
- 1:27:13
- Unidentified speaker (speaker_3): Yeah.
- 1:27:14
- Unidentified speaker (speaker_2): You
- 1:27:15
- Unidentified speaker (speaker_3): Um unfortunately I uh had some operations on my vocal cords, which
- 1:27:19
- Unidentified speaker (UNKNOWN): means
- 1:27:19
- Unidentified speaker (speaker_3): it's really difficult to do um the audiobook for so long. So unfortunately it's somebody else. But uh But
- 1:27:24
- Unidentified speaker (speaker_2): I'm sure the AI can now replicate your voice.
- 1:27:27
- Unidentified speaker (speaker_3): I have done that actually. I yeah, I gave them uh yeah, I did a uh a festival where it was just after the operations I'd had and um basically I I got I trained an A up AI voice up and it was amazing. So I just typed and it spoke. We
- 1:27:40
- Unidentified speaker (UNKNOWN): had to
- 1:27:40
- Unidentified speaker (speaker_2): we had to do a a an advert for the podcast and I was on holiday in Italy and we couldn't get the lighting set up. We created it in AI. We got we did all right, didn't we?
- 1:27:50
- Unidentified speaker (UNKNOWN): It's
- 1:27:50
- Unidentified speaker (speaker_1): like 75-80%.
- 1:27:52
- Unidentified speaker (UNKNOWN): Not ever
- 1:27:53
- Unidentified speaker (speaker_2): some people notice, not everyone noticed. I'm like, we're not far away from that. We're not far away from you and I not even have to them do this, but that would be sad because that
- 1:27:59
- Unidentified speaker (UNKNOWN): would be
- 1:27:59
- Unidentified speaker (speaker_3): sad.
- 1:27:59
- Unidentified speaker (speaker_2): I wouldn't have made
- 1:28:00
- Unidentified speaker (UNKNOWN): it. You know,
- 1:28:00
- Unidentified speaker (speaker_3): that's
- 1:28:01
- Unidentified speaker (UNKNOWN): like going
- 1:28:02
- Unidentified speaker (speaker_3): for a walk in the car. countryside and not actually doing the walk. Well,
- 1:28:05
- Unidentified speaker (speaker_2): listen, anyone listening, please go and check out Marcus's books. We'll put them in the show notes. Thank you for coming in, Marcus. Thank you everyone for listening. See you later.
- 1:28:24
- Unidentified speaker (speaker_0): Legendary has arrived in Alberta with BetMGM Casino. Sign up to unlock access to over 4,000 casino games, including exclusive slots you can't spin anywhere else. Survivors ready? The iconic TV show is now an online casino game where you can tap into island fun. BetMGM has also brought the
- 1:28:44
- Unidentified speaker (speaker_0): prices right: Game of Thrones, Family Feud, MGM Millions, and so many more exclusive casino games. to Alberta. Get Las Vegas right from your phone with live dealer games streaming from real tables. And every time you play, you can enjoy instant deposit options and fast withdrawals. BetMGM Casino and its legendary
- 1:29:04
- Unidentified speaker (speaker_0): library of games are available. Download the BetMGM app. See BetMGM.com MGM.com for terms. 18 plus Alberta only. Please play responsibly, subject to eligibility. requirements. If
- 1:29:16
- Unidentified speaker (speaker_1): gambling is affecting your mental health or well-being, 211 Alberta is here to help. Call or text 211 or visit ab.211.ca. BetMGM operates pursuant to an operating agreement with the Alberta Eye Gaming Corporation. Standard message and data rates may apply.
More from The Peter McCormack Show
#230 - Paul Collier - We Have Broken Capitalism
Episode discussion: Paul Collier argues UK capitalism is broken by centralization, citing growth failures and proposing devolution and land value tax
Speakers’ claims: Paul Collier argues that UK capitalism is failing due to excessive centralization in Whitehall, which has stripped local governments of power and funds, leading to stagnant per capita growth. He contrasts this with successful decentralized models in Germany and Denmark, advocating for devolution to regional economic units and a shift away from net-zero policies that he claims raise energy costs. Collier proposes taxing land appreciation to discourage zero-sum speculation and encourages business leaders to enter politics to replace ineffective civil servants. He frames Brexit as a regional revolt against London's dominance rather than an anti-European stance.
#227 - Gideon Rachman - How World War 3 Could Start by Accident
Episode discussion: Gideon Rachman warns of accidental WW3 via hybrid warfare, Iran escalation, and global fiscal/demographic crises.
Speakers’ claims: Gideon Rachman argues that World War III is likely to begin accidentally through hybrid warfare tactics rather than deliberate intent, citing risks from Russia, China, Iran, and North Korea. He details the dangers of miscalculation in the Middle East, specifically regarding US-Iran tensions and depleted missile defense stocks. Rachman also analyzes the domestic political constraints in Western nations, highlighting how strongman leaders prioritize legacy over governance. The discussion covers the fiscal traps facing European economies and the long-term economic threats posed by collapsing global fertility rates.