The HoaQ Podcast

Speakers analyze AI's economic impact on Africa, highlighting infrastructure gaps, data sovereignty opportunities, and investment risks.

52 min Listen to the original Episode page
Artificial IntelligenceAfrican Tech EcosystemData SovereigntyInfrastructureVenture CapitalGenerative AI

Summary

Dr. Chinasa Okolo and Cobih Obih discuss the state of AI in Africa, noting that while generative AI has democratized technical access, significant infrastructure barriers like expensive internet and unreliable power persist. They argue that African startups should avoid building 'wrapper' applications vulnerable to big tech consolidation, instead focusing on local data labeling, language models, and essential industry use cases in healthcare and agriculture. The speakers highlight Instadeep’s $700 million acquisition as a benchmark for success and emphasize the need for government support in compute infrastructure. They also warn of job displacement risks but suggest upskilling and human oversight as primary mitigation strategies.

Episode sections

  1. 0:00 – 4:58

    Introduction and Speaker Backgrounds

    Host introduces guests Dr. Chinasa Okolo (PhD in Computer Science, AI governance for African Union/Nigeria) and Cobih Obih (Product Manager at Stack Overflow). They discuss their experience as members of HOAC, having made ~30 investments each since 2020.

  2. 4:58 – 7:07

    Definition and Boom of Generative AI

    Dr. Okolo defines AI and identifies ChatGPT's late 2022/2023 launch as the catalyst for mainstream adoption. Cobih Obih notes ChatGPT forced Stack Overflow to pivot its strategy.

  3. 7:07 – 10:55

    Pros and Cons of AI

    Dr. Okolo highlights AI's potential in agriculture, healthcare, and education but warns of 'hallucinations' and environmental costs (16 ounces of water per prompt). Cobih cites AI failures like Google Gemini's initial image generation issues.

  4. 10:55 – 16:04

    AI Applications in African Industries

    Cobih discusses AI in healthcare (AlphaFold), finance, and agriculture (Digital Green). He emphasizes 'market-creating innovations' to solve non-consumption issues like personalized medicine and fraud detection.

  5. 16:04 – 19:01

    Economic Impact and Productivity in Africa

    Speakers discuss how AI boosts productivity for small businesses. Cobih notes AI democratizes technical knowledge, allowing non-experts to learn complex concepts like AWS quickly.

  6. 19:01 – 25:08

    Job Displacement Risks and Mitigation

    Dr. Okolo argues job displacement risks may be overstated, citing McKinsey and PwC reports. She emphasizes upskilling workers and data protection regulations for human oversight.

  7. 25:08 – 30:17

    Attracting Investment and Data Sovereignty

    Speakers advise African startups to focus on essential use cases and local data rather than building wrappers. They highlight opportunities in local language models (Lingpa) and data labeling (Amini, Awari).

  8. 30:17 – 31:30

    AI Strategy for Founders

    Dr. Okolo advises founders to target previously unsolvable problems using AI to reduce engineering team needs, citing Stack Overflow's use of AI for backlog items.

  9. 31:30 – 33:08

    Compute and Market Consolidation

    Speakers identify compute access and cost as primary barriers. Dr. Okolo predicts wrapper startups will fail as big tech integrates AI into their ecosystems, leading to market consolidation.

  10. 33:08 – 34:25

    Data Quality and Privacy Opportunities

    Dr. Okolo highlights the lack of clean data in Africa as an opportunity for local labeling services. She emphasizes GDPR compliance and suggests government support for compute infrastructure.

  11. 34:25 – 35:54

    Infrastructure and Skills Gaps

    Discussion covers infrastructure deficits, including expensive/slow internet in Nairobi vs. Dublin and unreliable electricity. Dr. Okolo calls for scaling AI skills training.

  12. 35:54 – 38:17

    African AI Success Stories and Grassroots Communities

    Instadeep is highlighted as a success story acquired for ~$700 million. Speakers praise grassroots communities like Masakane and Data Science Nigeria for advancing NLP.

  13. 38:17 – 41:06

    High-Need AI Projects in Africa

    Key areas include digitizing healthcare, consumer-facing tools, and domain-specific models for education/agriculture. Example: AI tools helping students score higher on the JAMB exam.

  14. 41:06 – 52:06

    Infrastructure Challenges and Tiny ML

    Dr. Okolo suggests government infrastructure investment and adoption of 'Tiny ML' for edge devices. Open-source models from Meta and Google help bridge resource gaps.

Assets discussed

AssetKindStanceContext
OpenAIprivate companyneutralAnd so I would say it definitely boomed within late 2022, more so like late 2023, honestly, with the introduction of ChatGPT's um OpenAI's Chat GPT tool.
AlphabetGOOGLequityneutralSo, you know, Google and Accra, IBM in Kenya, South Africa, Microsoft, um, what role do you see for these AI research labs in the market moving forward?
MicrosoftMSFTequityneutralSo, you know, Google and Accra, IBM in Kenya, South Africa, Microsoft, um, what role do you see for these AI research labs in the market moving forward?

“Stance” is how the speakers talked about the asset, not a recommendation.

Claims worth verifying

Dr. Chinasa Okolo finished her PhD in computer science at Quinnipiac University last year (relative to the 2026 podcast date).

Unidentified speaker ·0:41

How to verify: Check Dr. Chinasa Okolo's LinkedIn profile or academic publications for her PhD graduation date and institution.

Both speakers have made approximately 30 investments since joining HOAC around 2020.

Unidentified speaker ·2:25:16

How to verify: This is a self-reported personal claim; verification would require access to the speakers' private investment portfolios or public disclosures if they are registered investment advisors.

OpenAI's ChatGPT was introduced in late 2022, with a major boom in late 2023.

Unidentified speaker ·5:50

How to verify: Check public records of OpenAI's product launch dates; ChatGPT was publicly released in November 2022.

Estimates suggest it takes about 16 ounces of water to prompt an AI system.

Unidentified speaker ·8:26

How to verify: Search for academic studies or reports from organizations like the International Energy Agency (IEA) or specific tech sustainability reports that quantify the water footprint of AI inference.

Google announced 'prompt caching' technology that reduces energy usage by 70%.

Unidentified speaker ·10:15

How to verify: Search for Google Cloud or Google AI press releases and technical blogs regarding 'prompt caching' and its claimed energy efficiency metrics.

Google DeepMind's AlphaFold was a major innovation in drug discovery before AI went mainstream.

Unidentified speaker ·11:32

How to verify: Check scientific literature and news archives for DeepMind's AlphaFold releases and their impact on protein structure prediction.

Kenyan workers were involved in training ChatGPT, performing data labeling and annotation, and were exposed to graphic content with insufficient compensation.

Unidentified speaker ·24:03

How to verify: Search for news reports, specifically referencing Time magazine or other investigative journalism, regarding OpenAI's use of Kenyan contractors for data labeling.

The Stanford AI Index Report showed that Kenya has a very high adoption rate of tools like ChatGPT.

Unidentified speaker ·23:33

How to verify: Access the latest Stanford AI Index Report and check the section on global AI adoption rates by country.

McKinsey and PwC reports suggest AI job displacement risks are overstated and that AI primarily upskills white-collar work.

Unidentified speaker ·19:37

How to verify: Search for McKinsey Global Institute and PwC reports on the future of work and AI's impact on employment, specifically looking for conclusions on upskilling vs. displacement.

Digital Green is a nonprofit using generative AI to democratize farming practices in Kenya and Nigeria.

Unidentified speaker ·13:10

How to verify: Visit the Digital Green website or search for their case studies and reports on AI-driven agricultural extension services in East and West Africa.

Stack Overflow used AI to solve 20 backlog items that previously required many engineers.

Unidentified speaker ·30:43

How to verify: Search Stack Overflow engineering blogs or technical case studies from the period when they integrated AI tools into their development workflow.

Apple Intelligence caused many AI startups to die.

Unidentified speaker ·31:59

How to verify: Review tech news reports and startup funding data from late 2024/2025 regarding the impact of Apple's AI announcements on wrapper startups.

Internet in Nairobi costs $40-$50/month for 20MB, while in Dublin it costs the same for 500MB.

Unidentified speaker ·34:31

How to verify: Compare current ISP pricing plans in Nairobi, Kenya, and Dublin, Ireland, for comparable data allowances.

Instadeep was acquired for around $700 million.

Unidentified speaker ·36:01

How to verify: Check press releases from Instadeep or its acquirer (DeepMind) regarding the acquisition price and date.

Instadeep helped develop one of the earliest COVID-19 vaccines.

Unidentified speaker ·36:31

How to verify: Search for Instadeep’s partnership announcements with vaccine developers (e.g., CureVac or others) during the 2020-2021 period.

The Nigerian government received significant investment from the World Bank for telecommunications infrastructure.

Unidentified speaker ·42:00

How to verify: Search World Bank project databases for loans or grants allocated to Nigeria for telecom infrastructure development.

Google has a flood forecasting system successful in predicting floods throughout Africa and worldwide.

Unidentified speaker ·47:27

How to verify: Review Google.org or Google AI blog posts regarding their flood forecasting initiatives and their geographic coverage.

IBM has AI research labs in Kenya and South Africa.

Unidentified speaker ·46:42

How to verify: Check IBM’s official website for locations of their research centers or offices in Kenya and South Africa.

Fondamate uses GPT-3 and has a partnership with Meta to leverage Llama.

Unidentified speaker ·40:34

How to verify: Search for Fondamate’s technical blog posts, press releases, or GitHub repositories detailing their model usage and partnerships.

L'Elappe AI was launched by founders of Masakane to develop Africa's primary large language model.

Unidentified speaker ·37:34

How to verify: Search for L'Elappe AI’s launch announcements and verify the founders' previous affiliation with the Masakane community.

Do your own research

  • Verify the specific environmental impact metrics cited by Dr. Okolo regarding water consumption per AI prompt through independent sustainability reports from major cloud providers.
  • Review McKinsey and PwC reports on AI's impact on employment to assess the validity of the claim that AI upskills rather than replaces white-collar workers.
  • Research current pricing structures for internet bandwidth and cloud compute services in Nairobi versus Dublin to confirm the cost disparity claims.
  • Investigate the business models of local data labeling companies mentioned (e.g., Amini in Kenya, Awari in Nigeria) to assess their market viability and client base.
  • Analyze the competitive landscape for 'wrapper' AI startups to determine which specific features are being integrated by big tech firms (Apple, Google, OpenAI).
  • Examine the acquisition details of Instadeep to understand the valuation drivers and post-acquisition performance.
  • Check for recent government initiatives in the Middle East (e.g., AI-71) and Africa regarding compute infrastructure funding to evaluate the feasibility of similar models.
Full transcriptexpand
0:00
Unidentified speaker: You are listening to the Hulk podcast.
0:02
Unidentified speaker: Hulk is a community of creators and operators, parking entrepreneurs, building for Africa and its diaspora.
0:09
Unidentified speaker: Let's get into it.
0:10
Unidentified speaker: So today I'm super excited.
0:12
Unidentified speaker: Oh my god.
0:13
Unidentified speaker: I'm excited to, you know, to be to introduce our speakers for today.
0:18
Unidentified speaker: We have Chinasa and we have Kobe.
0:21
Unidentified speaker: I sort of do not want to introduce them.
0:23
Unidentified speaker: I want them to introduce themselves, right?
0:26
Unidentified speaker: So that, you know, I mean, I don't think anybody can introduce you more than you introduce yourself.
0:30
Unidentified speaker: So I would pass it on now to Tinasta to introduce herself and then to Kobe to introduce himself and we can get the conversation going.
0:39
Unidentified speaker: So hi everyone.
0:39
Unidentified speaker: I'm Chinasa Teocolo.
0:41
Unidentified speaker: I just finished my PhD last year in computer science at Quinal University.
0:45
Unidentified speaker: And now I work at the Burg Institution in Washington, DC, which is a public policy think tank.
0:50
Unidentified speaker: And so my work more broadly focuses on how African governments can best enable effective AI governance.
0:57
Unidentified speaker: And then also I do some work with the African Union and developing their AU AI continental strategy for Africa and also with the Nigerian government on developing their national AI strategy.
1:07
Unidentified speaker: So very happy to be here.
1:08
Unidentified speaker: Okay, thank you very much for that.
1:09
Unidentified speaker: Kobe, do you want to introduce yourself before we get into conversation?
1:15
Unidentified speaker: Yeah, of course, uh Chanessa is a hard uh hard active fellow.
1:19
Unidentified speaker: I'm just a lowly product manager who've worked in uh several big tech companies over the last decade.
1:25
Unidentified speaker: And uh in my last role, I was building and leading uh product for Stack Overflow Web beyond generative AI uh solutions to solve problems that our users were having and using AI to solve those everything problems that we couldn't solve previously.
1:40
Unidentified speaker: Uh and it's you know it's an honor to be on the same stage as Junessa because she's an incredible human being.
1:47
Unidentified speaker: So I'm happy to be uh sharing stage with her and then hopefully we're gonna learn more together from her knowledge.
1:52
Unidentified speaker: Thank you.
1:54
Unidentified speaker: Okay, great, great.
1:55
Unidentified speaker: Um, I know like this was not our plan, but maybe, you know, before we get started, you can just talk a little bit about your.
2:03
Unidentified speaker: I know both of you are like OG Oak members, right?
2:06
Unidentified speaker: Maybe you just talk about your time at Oak, a little bit about your investments, and then we can like go into conversations.
2:14
Unidentified speaker: Is that okay?
2:15
Unidentified speaker: Yeah, that's fine.
2:15
Unidentified speaker: I'll start.
2:16
Unidentified speaker: So yeah, I've been a member of HOAC for about four years now, and so have gotten a really good opportunity to invest in companies across the African continent and also in the diaspora.
2:25
Unidentified speaker: And so I think I've made maybe close to 30 investments or so over the past couple of years, which has been really great, and been also fortunate to see some exit exits as well.
2:34
Unidentified speaker: And so definitely looking forward to continuing my journey with HOA.
2:37
Unidentified speaker: But yeah, the um I felt for me that um at the time in my life when I started investing in with HOG that I just wanted to diversify and also get an opportunity to help support some of the emerging tech startups within the continent and also the diaspora.
2:52
Unidentified speaker: And so it's been really great to see how um my profiling companies and also just HOA's companies have evolved over the past couple years, and also great to see how um the investment trends have changed over time, especially um due to the impact of the COVID pandemic.
3:08
Unidentified speaker: Similar to Chinesa, like I joined HOAC about uh 2020, so about four years now, and I've made about 30 plus investments in that time as well.
3:16
Unidentified speaker: Um, like uh her have also had some some small exit, or hopefully we get some big exit.
3:23
Unidentified speaker: We know what has happened to the market in the last couple of years, so uh the market's a little you know contracted at the moment, but hoping for rebounds uh towards the end of this year to uh next year.
3:34
Unidentified speaker: Uh but you know it's always down to sports.
3:37
Unidentified speaker: Um founders who are doing great things, improving the lives of uh people on the continent.
3:45
Unidentified speaker: With what we're talking about today, uh AI, this is gonna be incredibly important.
3:50
Unidentified speaker: How founders are gonna be tapping these new technologies to even accelerate other progress they're doing or find new problems to solve that they couldn't solve before.
3:59
Unidentified speaker: So hopefully this conversation we'll have today is gonna uh eliminate some of those ideas that people can start thinking about and take forward.
4:06
Unidentified speaker: And obviously, if you're a founder and you decide to do that, please make sure you invite Hulk because we'll love to invest in any uh generative or AI solutions that you're looking at to build for for the continent.
4:19
Unidentified speaker: Thank you.
4:20
Unidentified speaker: Okay, great, great, great.
4:22
Unidentified speaker: Um, just before they start, please, if you have any questions, feel free to just raise up your hands or you can just use the chat.
4:28
Unidentified speaker: And with that being said, I'll pass it on to Kobe and Chinasa to educate us, really.
4:36
Unidentified speaker: All right, I'm gonna uh again, it's an honor to share the stage with Chinaser.
4:40
Unidentified speaker: Uh, you're a very accomplished human being.
4:43
Unidentified speaker: Uh so happy to have you here.
4:45
Unidentified speaker: Uh today I'm gonna try to uh learn more along with all the people who are viewing us online uh around AI.
4:52
Unidentified speaker: I only know a little bit, but that's your field, you know.
4:56
Unidentified speaker: So uh let's start with some live questions.
4:58
Unidentified speaker: What is AI and what has happened in the last uh year or so in the AI space?
5:04
Unidentified speaker: Yeah, definitely.
5:05
Unidentified speaker: So yeah, thanks, Hobby.
5:06
Unidentified speaker: Also I'm an honor to be share the stage with you for this AMA and looking forward to the discussion that we have.
5:12
Unidentified speaker: So uh to get started about basically a very simple definition of AI, we can consider it to be um, you know, different systems, especially those that are technologically um enabled to help them understand how to perform um complex tasks like you know, reasoning, decision making, and also solving different problems in various fields.
5:32
Unidentified speaker: And so to get to this, to get to this process, um these systems are trained on basically multitudes of data within specific fields that they could be um, you know, like image data, voice data, um, or you know, text data, which has been very popular for different applications.
5:50
Unidentified speaker: And so I would say it definitely boomed within late 2022, more so like late 2023, honestly, with the introduction of ChatGPT's um OpenAI's Chat GPT tool.
6:01
Unidentified speaker: I feel like this is the first time that you know people that were not technologists or computer science researchers or just like those heavily involved in tech fill were actually able to interact with AI for the first time and really understand like with its power.
6:16
Unidentified speaker: And so I definitely think that, you know, if open AI kind of set the precedent for what we know as AI today.
6:21
Unidentified speaker: We also know that you know generative AI is only a very small subset of what AI is overall.
6:26
Unidentified speaker: And so there's many other different aspects of it, like computer vision, um deep learning, natural language processing, et cetera.
6:33
Unidentified speaker: Um, and so yeah, we'll probably get into this more in a conversation.
6:37
Unidentified speaker: Thank you for that.
6:38
Unidentified speaker: I I agree.
6:39
Unidentified speaker: Like I was working as Stack Over for building something unrelated to AI, then ChatGPT dropped, and I went my entire strategy got blown apart, and we had to pivot the entire company like response to that.
6:52
Unidentified speaker: So I can agree with that with the launch of ChatGPT, the AI space had been top of mind on and mainstream across users before it was just niche around, you know, people like yourself, I imagine, but now everyone knows Chat GPT.
7:05
Unidentified speaker: So uh thanks for giving us that background.
7:07
Unidentified speaker: Um, from your perspective, what would you say are the pros and cons of AI?
7:12
Unidentified speaker: Yeah, so for me, I would say the pros are definitely, you know, the ability for it to be applied to many different fields, especially within more pressing domains like agriculture, healthcare, education, and also I would say just the uh ability of AI overall to um, you know, understand and get um especially nuances from large swaths of data in order to make decisions.
7:34
Unidentified speaker: I think this also comes with some cons, especially in terms of how these these decisions um are being made.
7:40
Unidentified speaker: Sometimes I think that anyone who's interacted with um AI systems, especially uh chatbots, very extensively, understand that there are, you know, a large potential or high potential for these tools to let's say like hallucinate or create basically false information.
7:56
Unidentified speaker: I've experienced this myself when trying to outline ideas for research projects.
8:01
Unidentified speaker: And so um I think that this is a big issue that you know it's been hard for companies like uh OpenAI and Google to address in their chat box.
8:09
Unidentified speaker: And so hopefully we see some, I don't know, kind of redress for this soon.
8:13
Unidentified speaker: And also a big con that people don't necessarily are not necessarily aware of is that, you know, actually developing and using AI systems and tools takes up a lot of resources and is contributing to climate change.
8:26
Unidentified speaker: Um, companies like Microsoft, Google, and AI are investing heavily in data centers to actually power their systems and tools, and they're putting these um centers in deserts, which already you know are very sparse areas, and uh uh citizens and residents in these areas are having to fight back because it takes about there's different estimates in terms of like how much energy it takes to prompt an AI system, but estimates have to say like it takes about 16 ounces of water, and we know that millions of people are prompting every minute, and so um I think this is a big concern that honestly has to shape how actually interact um with these tools as well.
9:04
Unidentified speaker: And then with the cons of AI, we've also been aware of the many issues of algorithmic bias that's been happening over the past couple of years.
9:11
Unidentified speaker: Uh, people like Joey Bilamini, um a former researcher at MIT, has really exposed many of these issues, especially as it pertains to facial recognition.
9:21
Unidentified speaker: And now that we're seeing AI being applied in more and more nefarious contexts like warfare, for example, um, I think that we have to really be concerned about how these tools are deployed and also used for um to affect or impact marginalized populations.
9:38
Unidentified speaker: Yeah, to add to that, um we've seen some of this stuff have gone mainstream, like when Google um provided Google Gemini with the image creation, where the image creation was very D and an eye focused, it couldn't generate any white person, so they had to shut it down till today that product is not back online.
9:59
Unidentified speaker: And similarly, when they provided the AI summarization, we can and they had this pattern as Reddit, where it's like, oh, to make your pizza, you can add glue to it, and like people started calling it out.
10:11
Unidentified speaker: Um, but another con you mentioned is around the energy consumption.
10:15
Unidentified speaker: Uh, you notice that Google today announced that they now have this thing called prompt caching, where it would reduce the uh energy usage by 70%.
10:25
Unidentified speaker: That left to be seen if that's going to be effective, but they're trying to address some of these concerns around the cons of uh having these large language models and these high computes and these data centers across the world.
10:36
Unidentified speaker: Thank you.
10:37
Unidentified speaker: Yeah, this is really important because we also see that you know African countries bear the brunt of climate change while only contributing a very small number or like percentage of emissions.
10:48
Unidentified speaker: And so I think these initiatives will be very important and I hope to see them scale.
10:52
Unidentified speaker: So now I'm gonna get into asking um you, Kobe, some questions.
10:55
Unidentified speaker: You've had lots of experience um in tech, and would love to learn your experience, um, your insights on you know which industries are benefiting from AI now, and then also how can that apply to the continent?
11:07
Unidentified speaker: Thank you for that.
11:08
Unidentified speaker: I think uh in all the industries, really, you can the good the benefits of AI actually is very generalized.
11:15
Unidentified speaker: Like you can use them in different applications.
11:17
Unidentified speaker: Uh in healthcare, you can see uh AI being used in diagnostics by helping doctors diagnose uh x-rays and uh cancer applications.
11:29
Unidentified speaker: So AI can accelerate a lot of things that people are already doing.
11:32
Unidentified speaker: Uh in drug discovery, we see Google DeepMind do alpha folds, one of the biggest things that before AI went mainstream, that was the biggest uh innovation.
11:42
Unidentified speaker: I was at Google when they happen, and I can see like all the joy of what AI has accomplished there.
11:48
Unidentified speaker: Uh, personalized medicine, a lot of the in Africa we don't have as much specialized care as we have in the West.
11:56
Unidentified speaker: So AI can help accelerate and democratize the personalized medicine uh with patient care management.
12:02
Unidentified speaker: In finance, we AI again can be using risk management.
12:05
Unidentified speaker: There's a lot of investments going into that space.
12:08
Unidentified speaker: As we've seen, like the the CEO Pay Hippo, you know, who retired or who quit his job several months ago, starting a company in the risk and uh detection.
12:20
Unidentified speaker: And I think that's due to the proliferation of AI.
12:23
Unidentified speaker: There's another company called Aukre, based out of South Africa, they just got fundraised for half a million dollars, and they only started in January.
12:31
Unidentified speaker: So these are some tools that already exist in the West.
12:35
Unidentified speaker: So AI can help us do that.
12:37
Unidentified speaker: I believe in market creating innovations, you know, from the uh Prosperity Paradox book.
12:44
Unidentified speaker: You know, we have to look at what we how we adapt this technology by, you know, solving non-consumption.
12:51
Unidentified speaker: A lot of the challenges we have in Africa, it's not like the the technologies don't exist, they do, but we probably can afford it, or we done the infrastructure for it.
12:59
Unidentified speaker: I think AI can help us accelerate the most market uh creating innovations.
13:04
Unidentified speaker: So I think that's how we should look at it.
13:06
Unidentified speaker: And again, in agriculture, which is one of the big things we have to feed our people.
13:10
Unidentified speaker: Uh, there's this company called Digital Green, it's a nonprofit, and what they do is they generate AI to democratize uh good farming practices to farmers in Kenya and in Nigeria, and they're trying to like localize it in different languages so that way they can connect better.
13:28
Unidentified speaker: Um in education, Chat GPT you know, caused a big stir around people being able to like do their homework and now AI detectors.
13:37
Unidentified speaker: For me personally, when it comes to education, I can I use uh these generative AI to learn technical uh uh knowledge really quickly, like how you know what is AWS, what is EC2?
13:50
Unidentified speaker: Like these are not things uh I know off the head, but I in in a matter of hours, I can be, you know, maybe college-level understanding of these technical uh concepts.
14:01
Unidentified speaker: So AI can also support um education.
14:05
Unidentified speaker: In Africa, what we need to do is we need to start building local capacity.
14:09
Unidentified speaker: You know, for example, the the work that you're doing with the Nigerian government is a good start, you know, trying to build some AI research.
14:16
Unidentified speaker: Uh, we need to start addressing data challenges.
14:19
Unidentified speaker: I think AI inherently is a data product, right?
14:23
Unidentified speaker: It's not really magic, it's just algebra and advanced algebra of data, you know.
14:29
Unidentified speaker: So, but we need clean data, it needs a data that's uh local data.
14:34
Unidentified speaker: For example, if you the problem we have today is everything online is from the West, it's all in English.
14:40
Unidentified speaker: There's nothing around our ancestry, our villages, our customs.
14:45
Unidentified speaker: So we need to provide this data, this clean data for us to create our own AI.
14:49
Unidentified speaker: Uh, we need a partner uh for success, you know, with international organizations, research institutions, and private companies, as you can see, like in, say, for example, in the Middle East, um, Microsoft has partnered with AI-71, you know, in Japan as well, Sakana just launched.
15:07
Unidentified speaker: So those are the kind of things we need to do for Nigeria and for Africa to you know, do the own data pipeline, then partner with big organizations.
15:14
Unidentified speaker: And we should all obviously focus on impact, right?
15:17
Unidentified speaker: Let's use this AI to solve problems that around ESG, you know, problems like poverty, that's the big concern, healthcare is a big one, and food security.
15:26
Unidentified speaker: Everyone's going crazy around the cost of tomatoes, right, right now in Nigeria.
15:30
Unidentified speaker: So perhaps we can use AI to solve some of those challenges.
15:33
Unidentified speaker: Um, and obviously, ethical considerations need to be made.
15:36
Unidentified speaker: Uh, we need to ensure that these AI is used for good and not for harm because the chance of it harming us and putting all of us out of a job is very high.
15:45
Unidentified speaker: So, those will be the things I think I see today and how we can apply it to the continent.
15:50
Unidentified speaker: Nice.
15:50
Unidentified speaker: Yeah, thanks, Kobe, for those really interesting insights.
15:52
Unidentified speaker: I actually learned about some companies I wasn't familiar with.
15:56
Unidentified speaker: And so we're really seeing, like, you know, the um investment ecosystem uh kind of heat up within the continent when it comes to AI startups.
16:04
Unidentified speaker: And so, how do you see um AI contributing to the economic growth of African countries in particular?
16:10
Unidentified speaker: I think it can boost our productivity and efficiency, to be honest.
16:14
Unidentified speaker: Uh, with AI, one of the things you've already seen in the West, which might end up being a problem for us uh uh in low cost of living areas, is the fact that with one person you can do more.
16:26
Unidentified speaker: So I think you know, if you're a small business, if you know how to use these AI products, you can hire less people.
16:32
Unidentified speaker: An example is take me and Joey, for example, were building something, and Joey's like, we need a website.
16:37
Unidentified speaker: I'm like, yeah, we do.
16:38
Unidentified speaker: And with the help of AI, we were able to build a website in five minutes.
16:42
Unidentified speaker: I'm like, Joey, tell us the mission, tell us the vision, what do you try?
16:45
Unidentified speaker: And we in it felt like magic to him.
16:47
Unidentified speaker: You know, that happened in five minutes.
16:48
Unidentified speaker: So think of small businesses on the continent, instead of hiring itself a developer or a designer to do that.
16:54
Unidentified speaker: With the help of AI, they can pull up their website in five minutes and be open for business.
16:59
Unidentified speaker: Uh, you know, service sector, customer service is a big problem we have, uh, especially for Nigeria.
17:05
Unidentified speaker: We're just bad at customer service.
17:07
Unidentified speaker: I don't care what anyone says, right?
17:09
Unidentified speaker: One of the good things about AI is it's very polite.
17:12
Unidentified speaker: You know, you can use AI now to build a chatbot, using your your road, um your documentation to answer those questions.
17:20
Unidentified speaker: So now you have an AI who can do customer service for you while you sleep.
17:25
Unidentified speaker: And again, again, more polite than the average Nigerian small business.
17:29
Unidentified speaker: So that could be a productivity boost and so and can even increase business for people.
17:34
Unidentified speaker: Uh in healthcare, we've already talked about some uh AI diagnostics.
17:38
Unidentified speaker: I would like to see more of that happen on the content.
17:41
Unidentified speaker: Financial services, as I say, risk detection, fraud management.
17:44
Unidentified speaker: Everyone who is an accompaniment, especially in Nigeria, know how bad you know these things, these scams can be.
17:51
Unidentified speaker: So using this for detection, the AI now empower uh powers is something that I think we should see more of.
17:58
Unidentified speaker: Education is always a big one.
18:00
Unidentified speaker: You know, perhaps there's ways for AI to transcript.
18:03
Unidentified speaker: Actually, it's already happening.
18:04
Unidentified speaker: So something I was looking at the other day.
18:07
Unidentified speaker: Um, there was this program that I watched as a child that I felt like made me very techy.
18:13
Unidentified speaker: So German programs, all in German.
18:14
Unidentified speaker: I'm like, I wish I can drop this stuff to English or to maybe Swahili or Uruguay so my kids can watch it.
18:22
Unidentified speaker: So with AI, I know those things are going to be uh available in the near future.
18:26
Unidentified speaker: So there's so many opportunities that I think we can use AI in the continent.
18:30
Unidentified speaker: And I think and I think founders should start looking at this stuff, like what are challenges, what are problems they have in your daily lives that now they can see if AI compute and fix that.
18:39
Unidentified speaker: And if AI doesn't exist and fix that, perhaps that's an opportunity for them to go fix.
18:44
Unidentified speaker: So there's so much, so much we can do.
18:46
Unidentified speaker: Uh, I I think the the the coast is very clear for Africans to look at what's a what what how can we use this AI to uh to change our lives for the better.
18:58
Unidentified speaker: Thank you.
19:01
Unidentified speaker: And let's switch gears.
19:02
Unidentified speaker: Now let's come back to you.
19:04
Unidentified speaker: Uh so what are some measures you think can be taken to mitigate the risk of job displacement due to AI?
19:11
Unidentified speaker: We've seen that that's the big concern, right?
19:14
Unidentified speaker: Because you can reduce the cost.
19:15
Unidentified speaker: And Africa is a low cost of leaving region.
19:18
Unidentified speaker: Uh, we've seen what's happening in India with the government also saying, you know, putting some kind of regulation around it.
19:25
Unidentified speaker: So, from your perspective, what can we do as Africans to mitigate these risks?
19:29
Unidentified speaker: Yeah, I would honestly say that, you know, there's been so many different reports, especially from McKinsey and you know, other consulting firms like PWC.
19:37
Unidentified speaker: And I feel like, you know, I would say some of these risks are actually a little bit overstated just because a lot of the benefits where we're seeing AI happen, especially when it comes to like jobs themselves, are actually upskilling uh white collar work.
19:51
Unidentified speaker: And there's little impact, I would say, in some of the more traditional fields like factory work and manufacturing and you know, those are those areas.
19:59
Unidentified speaker: I think that there are many potential opportunities for African father specifically to contribute and also African research academic labs to contribute towards building tools and systems that help farmers, for example, um, you know, improve their crop yields or also help, you know, um factory workers improve how they build or develop specific tools.
20:21
Unidentified speaker: But I think the biggest benefit I see right now is for you know AI to help blue uh white color workers, you know, improve their respective output.
20:31
Unidentified speaker: So, for example, for academic researchers helping them to do better literature reviews to a certain extent, and then also produce higher quality um, you know, research publications, which can definitely help improve how funding um comes to these researchers.
20:48
Unidentified speaker: When it comes to mitigating the job risk or the risk of job displacement due to AI, I think the most important thing is upskilling workers to understand the limitations and benefits of AI.
20:59
Unidentified speaker: I think that again, just due to the many misconceptions about how AI works, I think that people feel like it can address basically anything that they want it to do without really understanding like how it can do so.
21:10
Unidentified speaker: And so I think that also really understanding like these limitations of AI can also help a manager specifically understand that it may not necessarily be able, you know, for AI to do this certain function and that a human will always have to be involved in this process.
21:27
Unidentified speaker: I think that the for many and also many existing regulations, especially data protection regulations in the continent, also provide um, you know, consumers with the ability to opt out of having their information processed through, you know, algorithms.
21:41
Unidentified speaker: And so humans will always have to have some oversight over that process.
21:45
Unidentified speaker: And I think, you know, upskilling is really important just because in Africa, particularly, you know, um residents, I always say suffer from a lack of digital literacy.
21:53
Unidentified speaker: And this also plays into how AI literate they are.
21:56
Unidentified speaker: And so we'll have to do a lot of work in ensuring that workers know about AI tools and then also that they're able to effectively leverage these tools in their work to make sure that they can use these tools in ways that, you know, don't engender um, you know, exist existing biases and also don't cause harm to marginalized populations.
22:16
Unidentified speaker: So to dig in a little bit deeper, uh when I I hear you around upskilling, I do think it's a big part.
22:24
Unidentified speaker: I mean, Africa's always needed to upskill uh one of the reasons where it'll compete today is the fact that a lot of people now we have a lot of software engineers, you know, now we can now that you know, world class, and now we can explore people.
22:36
Unidentified speaker: So I think for us to uh attract more uh work to the continent, we do need to upskill more people, we need to educate more people, we need to obviously the work you guys are doing uh uh with the Nigerian government is gonna you know work towards empowering more people.
22:51
Unidentified speaker: But I I also wonder is there nothing, you know, do you not worry that the the risk might be so severe?
23:00
Unidentified speaker: Because look at the people who are leading this work.
23:02
Unidentified speaker: We have some Altman who is like the most microbiant CEO that we've seen in the last uh decade or so.
23:11
Unidentified speaker: Even when you have people like Elon Musk worried about you, I think the rest of us should be concerned.
23:17
Unidentified speaker: So, do you not think there's that government should should take a closer look and maybe put some guardrails around these technologies to make sure that we don't end up being beholden to some Americans who control the entire AI world?
23:33
Unidentified speaker: Yeah, I think this is really important just because, you know, for example, the Stanford AI index report showed that Kidney's actually have a very high adoption rate of tools like ChatGPT.
23:43
Unidentified speaker: And so these tools are basically infiltrating African countries and being used on a daily basis by African consumers.
23:50
Unidentified speaker: And so I think it's really important that they um, you know, be, I would say, supervised, and also that for you know any, let's say, misgivings that happen, that you know, countries themselves are actually able to seek recourse.
24:03
Unidentified speaker: So, for example, um, you know, I'm not sure if many people know this, that um time reported on how many Kenyan workers were actually involved in the training of Chat GPT itself.
24:14
Unidentified speaker: And they were very involved in the, I would say, the lower tenants of this, which is data labeling and data annotation.
24:21
Unidentified speaker: However, they were really repeatedly exposed to graphic content, they also worked very long hours and were not properly or sufficiently compensated for their work.
24:30
Unidentified speaker: And so there's been a lawsuit from these workers against open AI, but just given you know the respective um enforcement capacity and also legal capacity of African countries or Kenya specifically, it's really hard to understand if these workers will ever seek recourse from AI.
24:46
Unidentified speaker: So I think that it's important that we bolster our existing um protections throughout the throughout the African continent to ensure that these kinds of harms don't continue.
24:56
Unidentified speaker: All right.
24:57
Unidentified speaker: Um so, Michelle, I know you have your hands up.
25:00
Unidentified speaker: If you don't mind, we we're gonna wait to the end to ask questions.
25:03
Unidentified speaker: Is that okay?
25:06
Unidentified speaker: All right, thank you for that.
25:08
Unidentified speaker: All right, so one more question for you, folks who scares.
25:11
Unidentified speaker: How can Africans try to attract more investment in AI technologies?
25:16
Unidentified speaker: Yeah, so one thing I think is really important is really understanding where to provide value within the AI ecosystem.
25:22
Unidentified speaker: I think that, you know, a lot of startups after you know the first major two releases of ChatGPT were really focused on providing wrappers for basic functions for you know OpenAI's tools and products and leveraging their API.
25:35
Unidentified speaker: But we saw how in subsequent releases that their essential product market was obliterated by these updates.
25:42
Unidentified speaker: And so I think that it's important for um, you know, African startups to really focus on, I would say, essential um use cases of AI.
25:50
Unidentified speaker: And then also I think that I think the data and the models themselves are very valuable.
25:54
Unidentified speaker: And so providing ways, you know, for novel models that necessarily are able to understand African data and context.
26:01
Unidentified speaker: And I think this is really important just because we also see that the majority of these popular AI tools on the market, they don't serve African languages.
26:09
Unidentified speaker: And if they do, they don't serve them well.
26:11
Unidentified speaker: And so I think that there's a really strong need for, you know, like an African kind of chat GPP.
26:17
Unidentified speaker: I know this is already in process or in the in progress by companies like Lenapa and AI in South Africa that are focusing on you know targeting the South African languages, but they hope to expand to other languages as well.
26:30
Unidentified speaker: And so I think this is really important.
26:31
Unidentified speaker: And then also other startups like um uh Amini and Kenya and also Awari and Nigeria are really focused on the data side of things first and you know, providing this data that's needed to build and um, you know, develop these systems and also just providing, I would say, um also focusing on like the um labor aspect of it and providing opportunities for you know residents within these countries to participate meaningfully in this AI development process.
26:59
Unidentified speaker: And so that's where I see um how African startups can uh really attract significant investment for AI and then also hopefully gain significant product market fit for their tools.
27:12
Unidentified speaker: Yeah, sounds good.
27:13
Unidentified speaker: I completely agree.
27:14
Unidentified speaker: I mean, we've seen investments from uh these big AI companies like OpenAI and Google licensing data from like Reddit and Stack Overflow.
27:24
Unidentified speaker: Why not?
27:25
Unidentified speaker: Uh what's the popular Nigerian websites?
27:28
Unidentified speaker: Is it was it Naira, Naira Land?
27:30
Unidentified speaker: Yes, that's you know, it's in data on Reddit.
27:33
Unidentified speaker: You know, why is open AI should be paying them money uh to license that data?
27:37
Unidentified speaker: Um speaking of languages, like I said, Digital Green is one of those NGOs that using generated AI to democratize information of farmers.
27:46
Unidentified speaker: One of the challenges, my friend works as a director of product, and one of the challenges is like beyond Swahili, they don't they don't have enough data to uh get other languages like Hikuyu, because Swahili is just one language that's spoken in Kenya.
28:00
Unidentified speaker: So let's say you go to Nigeria, you might be able to get Yoruba or Ibo or Hausa, what about Urubo?
28:05
Unidentified speaker: You know, so some so there's a lot of opportunities for us to you know create new languages and sell this to AI companies or use the Buddha on AI chat GPT, for example.
28:18
Unidentified speaker: Like if you look at Duolingo, that's been around for many years, I've been waiting, still waiting for my language to be on there so I can I can refresh.
28:26
Unidentified speaker: So there is a huge opportunity for for Africans to attract investments there if we can solve problems that we are better suited to solve and uniquely uh empowered to solve.
28:38
Unidentified speaker: Thank you for that.
28:39
Unidentified speaker: Great.
28:39
Unidentified speaker: Yeah, thanks.
28:40
Unidentified speaker: And so we know that you know there are many opportunities to enter into AI and people maybe trying to understand how they can best position themselves for um to enter in ARO potentially.
28:50
Unidentified speaker: And so for you, Kobe, as someone who works in AI, uh what are the opportunities in the space and also how can people maximize them?
28:58
Unidentified speaker: Um there are a ton of opportunities in this space, to be honest.
29:02
Unidentified speaker: Like um some of the opportunities I just mentioned, right?
29:06
Unidentified speaker: You know, Naira Land is our Reddit, is being it's the only Reddit most Nigerians know.
29:12
Unidentified speaker: Granted, PPC is accessible, but people say the same thing about Reddit as well.
29:16
Unidentified speaker: So there's no real difference.
29:17
Unidentified speaker: So to me, that that's one opportunity now.
29:20
Unidentified speaker: Uh the other opportunities I mentioned, like our local languages, like those are dying.
29:25
Unidentified speaker: Like I'm Egbo, and for I don't speak Igbo fluently.
29:29
Unidentified speaker: And the reason is because I left Nigeria at an early age and I haven't had the tools to be able to educate myself.
29:34
Unidentified speaker: To me, that is that is a very huge opportunity.
29:37
Unidentified speaker: If someone creates in the Duolingo camp uh alternative that that I can learn Ebo, perhaps you know that's a that's a market right there because Duolingo's shown they they weren't willing to invest in that space.
29:49
Unidentified speaker: And there's a lot of African languages, so you can even have to worry about these Western Western companies because it will never be a big enough market for them to invest into doing it.
29:59
Unidentified speaker: So we have to be the ones that use this technology to invest in those spaces and solve those problems.
30:04
Unidentified speaker: Uh so like I said, so many opportunities.
30:06
Unidentified speaker: We've talked about the ones in healthcare, we've talked about uh the ones in uh finance, uh, we've talked about ones in customer service, now we're talking about the ones in language learning.
30:17
Unidentified speaker: So again, using this market creative innovations um strategy, we should be looking at problems that perhaps due to non-consumption or price uh or cost challenges or a lack of infrastructure.
30:33
Unidentified speaker: Now we can use AI to solve them.
30:36
Unidentified speaker: Those would be the things I would expect founders to be looking into.
30:39
Unidentified speaker: In fact, when I was a Stack Overflow, that's exactly the strategy we used, right?
30:43
Unidentified speaker: There were a lot of problems we challenges we had on the Stack Overflow website or the platform that we couldn't solve what would the AI is like, oh wow, we can solve these 20 things that was on our backlog that we needed so many engineers to solve.
30:56
Unidentified speaker: So those are the things I would expect founders to be looking at.
30:59
Unidentified speaker: What are problems they couldn't solve before due to maybe the cost to hire so many engineers and designers and data scientists?
31:06
Unidentified speaker: Now that cost gets reduced because you have a foundational model that can get you 80% there, and you need humans to get you the rest of the way.
31:14
Unidentified speaker: So that would be my challenge for any founders listening on this call.
31:18
Unidentified speaker: Any aspiring founder that's listening on this call, this is your call, call to action to go look for problems that you have in your daily life and see how AI can solve them for you.
31:27
Unidentified speaker: Most definitely, thank you.
31:30
Unidentified speaker: And so, next, um, what do you see as the main challenges related to access to AI technologies in Africa?
31:37
Unidentified speaker: Uh the biggest one would be compute, as you can see, like or AI when it started, you know, what November 2022, you know, but that's when Chat GPT dropped.
31:48
Unidentified speaker: A lot of people started putting AI wrappers around it, around, but now we've seen like those the startups are dying one by one.
31:56
Unidentified speaker: It's either OpenAI kills you or Apple kills you or Google kills you.
31:59
Unidentified speaker: I mean, Apple did with Apple Intelligence, whoop a lot of AI startups just died that way.
32:05
Unidentified speaker: And I expect that to keep happening.
32:06
Unidentified speaker: So, which means only the incumbents, the big guys who have with the computes, are really gonna be making the money.
32:12
Unidentified speaker: So I think um that's gonna be a huge challenge because these computers are not cheap, right?
32:16
Unidentified speaker: Because AWS is gonna say, oh, yeah, you want to do an AI, yes, we're gonna sell you, you know, access to AWS.
32:23
Unidentified speaker: But as we've already seen some cries earlier DC and last year, with African founders talking about AI, uh AWS is so expensive.
32:30
Unidentified speaker: So to me, I think that's gonna be one of the biggest challenges.
32:33
Unidentified speaker: And I also see that as an opportunity for African governments to support uh these enterprises.
32:38
Unidentified speaker: Like I said, in the Middle East, you see AI-71, even though it's a private, well, it's not really private companies, the government, you know, funding billions of dollars for these computes and these stuff for them to be able to compete with the world against Canada and Japan.
32:53
Unidentified speaker: So I think we need to do stuff like that, either as a country like Nigeria or as a region uh like ECOS or the entire continent, the African Union, of saying, hey, we're gonna have our own computes uh so that way we can empower these new uh founders.
33:08
Unidentified speaker: Again, data quality is a problem.
33:11
Unidentified speaker: We're allergic to data in the continent.
33:14
Unidentified speaker: It's kind of hard to find data on the continent.
33:17
Unidentified speaker: You know, it's like it's a kryptonite for us.
33:20
Unidentified speaker: We need to build uh a pipeline of our data and also find the ways to clean it.
33:25
Unidentified speaker: Because if you you're talking about how a lot of Kenyans were employed by OpenAI to do this stuff, you know, I think they employ scale AI who have labelers in Kenya.
33:36
Unidentified speaker: So we already have the capabilities of cleaning this data.
33:39
Unidentified speaker: So why not us, right?
33:42
Unidentified speaker: These Western companies are actually coming to us to do this data cleaning.
33:45
Unidentified speaker: So why don't we do this?
33:47
Unidentified speaker: Again, that's an opportunity for anyone listening to us.
33:49
Unidentified speaker: Open your own scale AI to start doing this data labeling annotation for African companies and selecting African currencies.
33:56
Unidentified speaker: That's another opportunity right there.
33:58
Unidentified speaker: Uh so uh data privacy concerns, we're not very good at protecting people's privacy.
34:05
Unidentified speaker: Again, if we're gonna compete in the world, we gotta use, you know, try to meet standards like GDPR, which is the best when it comes to data privacy, right?
34:12
Unidentified speaker: So we need to be thinking one with this software, beauty technologies, how do we make sure it's addressed, you know, GDPR?
34:18
Unidentified speaker: Because no company from the West is gonna buy that data if you know it's not compliant, because they still have to spend money to make it compliant again.
34:25
Unidentified speaker: Uh, we also have the infrastructure challenges like you know, limited connectivity.
34:30
Unidentified speaker: Internet is still very expensive.
34:31
Unidentified speaker: The amount of money I pay for uh for internet in Nairobi that I'm sitting, I pay maybe like $40 to $50 a month for 20 megs.
34:42
Unidentified speaker: That same amount of money in Dublin gets me 500 megs of data, you know, bandwidth.
34:47
Unidentified speaker: So it's like so we need to reduce the cost of data so that way more people can be connected to the uh like I say, inefficient computing power rate talked about the skills gap.
34:57
Unidentified speaker: You you mentioned that earlier, we need to help skill people.
35:00
Unidentified speaker: I know part of the work you do with the Nigerian government is trying to do that.
35:03
Unidentified speaker: We need to actually scale that up faster and get to the you know the places that need us the most.
35:09
Unidentified speaker: You know, so there's so many things we can do.
35:11
Unidentified speaker: Financial constraint, high cost, I've I've touched on that.
35:14
Unidentified speaker: And policy and regulation, again, I brought that up earlier on.
35:17
Unidentified speaker: Governments need to put guardrows around them.
35:19
Unidentified speaker: And uh socioeconomic factors, you talked about the cultural barriers that we need to you know uh uh fight through because in some parts of the world AI still feels like magic, which magic means witches, which is we need to burn them, we don't touch them.
35:33
Unidentified speaker: So there's there's certain things we need to do.
35:36
Unidentified speaker: So there's a lot of work to be done, and I think uh with a conservative effort, we can we can start inching away at these problems to make some progress.
35:43
Unidentified speaker: Agree.
35:44
Unidentified speaker: Yeah, you touched on so many different aspects, and so I hope that you know those aspiring founders in the room are taking note.
35:51
Unidentified speaker: Thank you.
35:52
Unidentified speaker: Uh so let's switch gears again.
35:54
Unidentified speaker: Uh, can you share any AI success stories from Africa that you know of?
35:58
Unidentified speaker: So I think, yeah, the most prominent one is Instadeep.
36:01
Unidentified speaker: And so I think a year or so ago they were acquired for um around 700 million.
36:06
Unidentified speaker: And so I think that's probably one of the most prominent exits within just the whole entire like a static ecosystem within Africa.
36:14
Unidentified speaker: And so would it be in it sets a really great precedent?
36:17
Unidentified speaker: Um, they're based in Tunisia.
36:18
Unidentified speaker: They also have offices um in Europe, I believe, and in France as well.
36:22
Unidentified speaker: And so um I think they've done really great work, one in just like publishing um in premier machine learning conferences, um especially those that are held in Western countries.
36:31
Unidentified speaker: And then also um they had a really interesting partnership and helping develop um one of the earliest COVID 19 vaccines.
36:38
Unidentified speaker: And so I think I think this is also what helped propose um, you know, their um their company and also show that and put them on a global stage.
36:46
Unidentified speaker: And so we'd really love to see more African countries, I see African startups, um, you know, have this kind of exposure that NCDEP has gotten.
36:54
Unidentified speaker: Also, I think that the grassroots AI community within Africa is honestly very thriving and uh point to them a lot because we've seen lots of different advances, especially uh with the machine translation and natural language processing or NLP.
37:07
Unidentified speaker: And so, for example, um some groups like um Masakane, um, Deep Learning in Daba, um, Data Science Nigeria, um, AI Saturdays Lagos, and Ghana NLP, um, and many more have been really focused on building these niche grassroots communities of AI researchers and developers to tackle and address African problems, especially as it pertains to preserving culture and language.
37:31
Unidentified speaker: And so I really love the work that they've done.
37:34
Unidentified speaker: Also, some of the founders of Masakane actually launched their own startup, L'Elappe AI, last year.
37:40
Unidentified speaker: I talked about them earlier in the call, and they focused on um, you know, developing Africa's primary large language model.
37:46
Unidentified speaker: And so we'd really love to see more of these spin-offs happen uh to improve uh the AI ecosystem overall.
37:53
Unidentified speaker: Yeah, it'd be nice.
37:54
Unidentified speaker: I I read about that InstaDeep razy, it was massive.
37:57
Unidentified speaker: Like it came out of nowhere, never heard of about them before.
38:00
Unidentified speaker: Then yeah, 700 million.
38:02
Unidentified speaker: Like that's that's a huge, huge, huge pay there.
38:04
Unidentified speaker: And I would believe like we need we need more Instant Deeps on the continent.
38:08
Unidentified speaker: You know, we need more people building AI from the ground up.
38:12
Unidentified speaker: Um, one more question uh before we turn it over to the audience.
38:17
Unidentified speaker: What types of AI projects are most needed in Africa right now?
38:22
Unidentified speaker: Oh, yeah, there's so many.
38:23
Unidentified speaker: And I think Kobe, you also talked about um a couple as well.
38:27
Unidentified speaker: I think honestly, just like I think data projects will be very helpful just because, again, we don't really see a lot of interoperability between um the countries themselves and also within existing um country, uh the countries themselves.
38:41
Unidentified speaker: I think that it's really important.
38:42
Unidentified speaker: And we also have like a lot of paper-based systems, especially in the healthcare field.
38:47
Unidentified speaker: And I think that companies focus on either digitizing um, you know, this data and also putting it, making it or into formats that are accessible to others and also able to be machine read will actually provide a lot of value and also really um help uncover some of the issues that have been underlying within different African countries and also help provide or help accelerate the development of novel tools and solutions to help address many of these present issues of healthcare as well.
39:17
Unidentified speaker: I think that also in general, just models or developing AI models that can are actually able to be commercialized and developed into consumer-facing tools like ChatGPT will actually be very helpful just because we're seeing how open AIs have been able to capitalize on this and actually get in significant investment from companies like Microsoft and you know, other significant investments like um between companies like Entropic and Amazon have also shown the power that these AI companies bring to these large giants, compete tech giants.
39:52
Unidentified speaker: And so we'd love to see more work done in this space.
39:54
Unidentified speaker: And I think that um will be focused on the domain-specific issues, um, you know, the prominent issues, healthcare, education, agriculture are all there.
40:04
Unidentified speaker: I think education, we've seen a lot of um help in this space with the companies developing chatbots and personalized tutoring tools for students.
40:12
Unidentified speaker: I just saw um a company, uh can't remember their name, unfortunately, talk about how their tool was actually helpful in students scoring high on JAM in Nigeria.
40:20
Unidentified speaker: And so would love to see more um efforts towards this.
40:24
Unidentified speaker: Yeah, those two is actually built by a friend uh in Dublin, Fondamate.
40:29
Unidentified speaker: Uh they've been building this work for before Chat GPT even dropped.
40:34
Unidentified speaker: You know, they use GPT-3, and and now I think they do have some partnership with uh Meta where they're gonna be leveraging Lama.
40:40
Unidentified speaker: So it's nice to see that that's you know resonating across the globe from beyond just our Dublin shores.
40:47
Unidentified speaker: Uh now we're gonna open up uh the the questions to the audience.
40:51
Unidentified speaker: Uh Michelle, you I you previously had your hand up.
40:54
Unidentified speaker: I I don't for like me, how much time do we have for questions?
40:57
Unidentified speaker: You know, I want to be mindful of the time here.
41:01
Unidentified speaker: Um another five, let's give it another five minutes, maybe.
41:05
Unidentified speaker: All right, in that case.
41:06
Unidentified speaker: Yeah, in that case, we're gonna take maybe three questions and we're gonna try to be very, very concise with our answers.
41:12
Unidentified speaker: Uh Michelle, you want to go first?
41:15
Unidentified speaker: Hi, thank you.
41:16
Unidentified speaker: This was so informative.
41:17
Unidentified speaker: Um so well done for having this.
41:21
Unidentified speaker: Um, I just want to know your opinion on if you're trying to go towards um AI in Africa, but there's still basic issues like light, light, especially when you talk about data, you know, storing data.
41:37
Unidentified speaker: I know that there's the cloud, but there's also on-premise that involves electricity, that involves things like that.
41:44
Unidentified speaker: Like, so how can you um how can Africa, how can Nigeria especially progress in AI, but they haven't even solved the light issue?
41:54
Unidentified speaker: Like, how will that work?
41:56
Unidentified speaker: I'm gonna let China take this one because he's working with the Nigerian government anyway.
42:00
Unidentified speaker: I think it's honestly like a tandem effort.
42:02
Unidentified speaker: You know, the the government has received significant investment from the World Bank to focus on develop uh expanding um, I would say, like uh telecommunications infrastructure.
42:11
Unidentified speaker: And I'm not sure of the progress of that, but and uh it would be great to see more of this happen, but it also has to happen like um an advanced speed.
42:21
Unidentified speaker: And also there are you know different methodologies within um AI that actually help people run and run and actually uh train models on smaller devices.
42:31
Unidentified speaker: So mobile computing and also like tiny machine learning or tiny ML has also been a field that's uh taken off, especially within like the Raspberry Pi community.
42:39
Unidentified speaker: And so I think that also it's actually not necessary to have a very large model to accomplish, you know, basic tasks.
42:46
Unidentified speaker: I think that people should also embrace um, you know, smaller models, especially for very niche use cases.
42:52
Unidentified speaker: And so um this can definitely help expand and how people are able to uh train and leverage these systems and also um you know, people the open source AI community has also been very uh helpful.
43:04
Unidentified speaker: You know, Meta just released some new models today.
43:07
Unidentified speaker: Um, and then also um Google has models and huggy phases have been a big repository for people to access, or researchers specifically to access models.
43:16
Unidentified speaker: And so um, this has kind of helped that gap.
43:19
Unidentified speaker: But you know, I would say countries definitely have to focus on this basic infrastructure for um African countries to be competitive within AI globally.
43:29
Unidentified speaker: Yeah, thank thank you for that uh question, Michelle.
43:32
Unidentified speaker: And I do agree with Chinasa, like, and I I like your question a lot, Michelle.
43:36
Unidentified speaker: Like when we start talking about how does this affect uh how do we you know compete in this age or competing with candles and and and lamps when our mates have uh solar energy?
43:49
Unidentified speaker: So, yes, African governments need to really take it seriously to empowers electrifiers, at least in the urban cores, if we're not talking about the villages.
43:59
Unidentified speaker: Maybe the cities should have you know high-speed uh internet, they should have 24-hour electricity, or make it cheaper for people to be able to pay their solar bills in installments.
44:10
Unidentified speaker: Like any country you go in the West, you don't have to pay your solar uh for your solar power right away.
44:16
Unidentified speaker: The government gives you stipends and so on.
44:19
Unidentified speaker: So there's so many things the governments can do to bring electricity, which is gonna in turn uh improve our productivity.
44:25
Unidentified speaker: Um, we're gonna go by the first to last culture.
44:28
Unidentified speaker: You have your Kobe, there's also a question for you in the chat as well.
44:32
Unidentified speaker: So maybe that first, yeah.
44:34
Unidentified speaker: So it's from a client, I believe.
44:37
Unidentified speaker: Uh Kobe, how how rarely would investors and VCs invest in startups not using AI in the next five years?
44:43
Unidentified speaker: Hi, that's a very tricky one.
44:45
Unidentified speaker: I like that question.
44:46
Unidentified speaker: I still think AI is still in the hype phase, right?
44:51
Unidentified speaker: I think if you ask, uh if you pick 10 random people on the street and ask them if they use ChatGPT in a day-to-day, you're gonna find that maybe only three of them do.
45:03
Unidentified speaker: So I I do believe that other things still, there's still a lot of other problems to solve, especially on the continent.
45:09
Unidentified speaker: AI is not gonna solve all the problems we have.
45:11
Unidentified speaker: Like Michelle just brought up one big problem we have in the continent, electricity.
45:17
Unidentified speaker: You know, you don't get AI without serving electricity, you know, high-speed internet.
45:22
Unidentified speaker: You don't get AI without serving high-speed internet, education.
45:25
Unidentified speaker: So there's so many problems the continent has that AI isn't the biggest problem we have, we just believe that AI can be a tool to accelerate some of the solutions that we prove, you know, we prefer.
45:36
Unidentified speaker: So I do think investors, smart investors and VCs should be investing in a lot of infrastructure projects.
45:43
Unidentified speaker: Because to me, those are the ones that are definitely gonna pay dividends for years to come.
45:47
Unidentified speaker: You know, so uh if a VC is not an investing style because there's no there's no AI in it, I feel sorry for that VC.
45:56
Unidentified speaker: But I do think in the next over the next five years, 10 years, when things a lot of things have improved, we're gonna see a lot of uh AI needs at that point.
46:04
Unidentified speaker: Um by that time, AI would probably either go mainstream or it would go the way of crypto where it's a boss.
46:08
Unidentified speaker: So that that the Jew is still out on that for for for for now.
46:13
Unidentified speaker: Um yeah.
46:16
Unidentified speaker: In the interest of time, not gonna China, I'm not gonna ask you to add your context because I would love to get to uh uh Kojo and do maybe so um happy to uh let's start with Kojo.
46:27
Unidentified speaker: Yes.
46:28
Unidentified speaker: Um hi Bob, thanks for the talk.
46:30
Unidentified speaker: Um really found it insightful and enjoyable, enjoyable.
46:33
Unidentified speaker: Um, I had a question really on um just to get your take on in terms of these multinational sort of AI research labs that we're seeing sort of pop up more in the continent.
46:42
Unidentified speaker: So, you know, Google and Accra, IBM in Kenya, South Africa, Microsoft, um, what role do you see for these AI research labs in the market moving forward?
46:51
Unidentified speaker: Um, and also in terms of how they operate working with local communities compared to like homegrown grassroots groups, uh you know, the LA LAPA and they've all been mentioned.
47:01
Unidentified speaker: But like what role do you see for these multinational AI research labs um in other context?
47:07
Unidentified speaker: Yeah, so um I actually had the chance to talk to someone about this earlier today, but I think that they do have a strong role to play in one, providing opportunities for African, you know, AI and machine learning researchers to um have opportunities to contribute to um products that will actually um you know be deployed and impact people's lives.
47:27
Unidentified speaker: I would say that Google um definitely has a really interesting like flood forecasting system that's been um really I would say successful in predicting floods throughout the continent and also worldwide.
47:36
Unidentified speaker: And I think like these applications are very much needed.
47:39
Unidentified speaker: Um IBM has done a lot of work um in healthcare, especially throughout the continent as well.
47:44
Unidentified speaker: Um I think that these um uh companies themselves also provide a launching pad for one, like people to start their careers and then also um hopefully to do like spin-offs or like go and venture off on their own to you know focus on you know their respective you know endeavors.
47:59
Unidentified speaker: And I think that this is a great opportunity um to like uh engage with these local communities themselves.
48:04
Unidentified speaker: I know that Google, for example, funds lots of different African academics at um you know research institutions.
48:10
Unidentified speaker: And I think that um this could definitely be uh increased and uh would love to see you know more um academic uh industry collaboration in this space Gobi, if you have any insights too.
48:23
Unidentified speaker: I know in the interest of time, and just we're just gonna keep rolling.
48:26
Unidentified speaker: Uh thanks for that question, Kojo, very good one.
48:29
Unidentified speaker: And do maybe is do maybe still with us?
48:31
Unidentified speaker: Yes.
48:32
Unidentified speaker: I am, she's still here.
48:33
Unidentified speaker: Um thank you so much.
48:34
Unidentified speaker: It was a pleasure listening to both of you.
48:36
Unidentified speaker: Um, particularly about that comment you made about investing in the AI value chain.
48:39
Unidentified speaker: I think that was pretty um, that's something we should look forward to.
48:42
Unidentified speaker: So my question is more so everyone is racing towards 2030, right?
48:46
Unidentified speaker: Um and and beyond uh kind of so my question is we are all trying, especially on the continent, to create the infrastructure and the policies that will enable us to be somewhat competitive, but from a pragmatic point of view, I wanted to hear your opinion.
49:02
Unidentified speaker: To in the instance that we are not able to measure up, um, what ways can we because what are your thoughts on ways we can position ourselves to maintain some semblance of independence or relevance in the conversation five, 10 years from now?
49:17
Unidentified speaker: I mean, I still believe like the way we should leverage our strengths.
49:22
Unidentified speaker: Our strengths are things that nobody else can do for us.
49:25
Unidentified speaker: You know, nobody else is gonna come to your village and you know, get that language and digitize them and label them and clean them so that way it can be used by some AI company.
49:36
Unidentified speaker: So those things are open AIs never come into your village.
49:40
Unidentified speaker: Open AI is never coming to my village.
49:42
Unidentified speaker: So there's all these things that these Western big conglomerates are never going to do.
49:46
Unidentified speaker: So there's things, those are things we can do, we can start doing today, we can keep doing.
49:50
Unidentified speaker: And whenever we do finally catch up, those things will still be useful.
49:54
Unidentified speaker: Uh, I think we should focus on our differentiators, what makes us different.
49:58
Unidentified speaker: There's certain things we shouldn't do.
50:00
Unidentified speaker: Let's not go be foundational models.
50:02
Unidentified speaker: One, it's too expensive, two, we don't have the money, we don't have it's there's a climate change effect to that.
50:08
Unidentified speaker: Let's do what we're definitely empowered to do, and we can do better than the other people.
50:14
Unidentified speaker: That would be my strategy for how Africans can still be competitive in that space.
50:19
Unidentified speaker: There are things most people can do and we need to do for ourselves.
50:23
Unidentified speaker: Over to you, Chinesa.
50:24
Unidentified speaker: Yeah, and and just to close up, I think that you know, investing and fundamental academic research is really important just because we've seen how this is the basic um, you know, tenet or the structure foundation for many popular advances in machine learning.
50:39
Unidentified speaker: I think that the US and Canada in particular have contributed heavily to this.
50:43
Unidentified speaker: And that and for you know, Africa to be competitive, we have to start publishing, you know, at these top AI and machine learning venues and um focus on improving our pipeline of uh masters and um PhD students um to ensure that they have voice um on the global stage of AI as well.
51:02
Unidentified speaker: Plus one.
51:03
Unidentified speaker: I appreciate that.
51:05
Unidentified speaker: Yeah, so we have gone way, way, way, way over time for like a me.
51:11
Unidentified speaker: No worries, no worries.
51:13
Unidentified speaker: I can tell, like, I mean, this was a great session.
51:16
Unidentified speaker: So, first off, I just want to thank the both of you for you know joining us here today and sharing your knowledge.
51:23
Unidentified speaker: And I'm sure everyone on this call also enjoyed, you know, the conversations.
51:26
Unidentified speaker: If you have any more questions for our speakers, feel free to just you know send us an email and we will pass it on to our speakers.
51:34
Unidentified speaker: And you can also, you know, follow them on LinkedIn, on Twitter, on Instagram, social media, really.
51:41
Unidentified speaker: So um Kobe Chinasa, please can you just drop your LinkedIn anywhere you want people to reach out to you?
51:47
Unidentified speaker: That would be great.
51:50
Unidentified speaker: Thank you.
51:50
Unidentified speaker: Thank you so much, everyone.
51:52
Unidentified speaker: Thanks for listening to the Hulk Podcast.
51:54
Unidentified speaker: If you enjoyed this episode, subscribe and follow the podcast for more conversations like this.
51:59
Unidentified speaker: To learn more about Hulk, visit Hulk.co and follow us across social media at Hulk Club.
This page was produced automatically: the audio was transcribed with nvidia/parakeet-tdt-0.6b-v3 and interpreted by Qwen3.6-35B-A3B-GGUF:UD-Q4_K_XL. Transcription and interpretation can contain errors; always confirm against the original episode. Nothing here is investment advice.
All insights