Cohere's Chief AI Officer, Joelle Pineau: Why Scaling Laws Will Continue & Future of Synthetic Data

Joelle Pineau 0:00 The scaling laws have been remarkably robust. There's a lot we don't know yet in terms of the vulnerability of these systems.

Harry Stebbings 0:07 Today, we have one of the leading minds in AI, Joelle Pino. Joelle is the chief scientist at Cohere. If you don't need to buy the Galacticos, why do you have, like, an Andrew Tulloch, a Daniel Gross, an Alex Huang, and the Galacticos assembling?

Joelle Pineau 0:21 I used to be quite skeptical that neural networks were necessarily the ultimate solution to machine learning. I seem to be quite wrong on this one.

Harry Stebbings 0:28 Knowing what you know, what do you not let your children do?

Joelle Pineau 0:33 Ah. (laughs) Eat too much sugar. I don't have a lot of patience as a scientist for people who are predicting the extremist scenarios, the catastrophic risks of AI. You know, AI becomes our overlord kind of scenario.

Harry Stebbings 0:50 If I gave you $10 billion, what would you spend it on first? Ready to go? (instrumental music plays) Joelle, it is so great to have you in the studio. I've heard many great things from Nick, Aidan, Schreppe. So, thank you so much for joining me.

Joelle Pineau 1:14 Thank you. Happy to be here.

Harry Stebbings 1:16 Now, you spent over six years at Meta, and I want to start there because it's a very transformative time and place. What are the biggest takeaways for you from that time? And how did that shape your mindset to how you think today?

Joelle Pineau 1:29 Well, I was there from t- 2017 to 2025. And you have to see just how much AI changed over that period of time. And what we were really focused on is fundamental AI research. Um, and what, you know, one thing that I've learned is just sometimes how long it takes to prove out a hypothesis. We feel like AI is moving at the speed of lightning. But in fact, there's some things that it just takes a few years to mature, to get the right optimizer, the right compute, the right data for that to really make a difference.

Harry Stebbings 2:02 I look at where we are today, and everyone kind of goes, "It's here, it's here, it's here."

Joelle Pineau 2:05 Yeah.

Harry Stebbings 2:06 And then you actually look at what a lot of the leaders have been saying recently, where it's like, actually, you know, um, Andrej was saying, "It's not the year of the agents, it's the decade of agents."

Joelle Pineau 2:14 (laughs)

Harry Stebbings 2:14 Sam's kind of pulling back too.

Joelle Pineau 2:16 Yeah.

Harry Stebbings 2:16 Have, have we got over our skis and we're actually kind of all pulling back, realizing that time is the factor we need

Harry Stebbings 2:22 to rely on?

Joelle Pineau 2:22 Well, I'll give you an example. You know, I've been in research for, for a couple decades now. I've been working on reinforcement learning for over 20 years. And suddenly, everyone's talking about reinforcement learning, you know? (laughs) Since the advent of reasoning models, agents and so on. So, you know, sometimes you have to be a little patient with these ideas. And the right algorithmic tweak, the right context, the right problem domains just opens up the magic.

Harry Stebbings 2:49 I, I was listening to Andrej yesterday, and he said in his show that reinforcement learning is terrible. Uh ...

Joelle Pineau 2:56 (laughs)

Harry Stebbings 2:56 (laughs)

Joelle Pineau 2:56 Less terrible than 20 years ago.

Harry Stebbings 2:58 Have we overinvested in RL-based methods at the expense of maybe, like, more scalable alternatives?

Joelle Pineau 3:05 Oh, I'm still super bullish on RL in that, like, the concept itself is so fundamental. You know, this idea of training through a system of rewards, of indicating what's valuable and what's not valuable through numerical values, like, that is so fundamental. It's not going away. Now, you know, where we're maybe getting a little bit ahead is thinking that just RL out of the box is going to give us AGI. That part, a lot less so. You know, if you look at the curve of progress, RL is terribly inefficient. And so the amount of signal you need to get in order to really shape the behavior of a model is far from where we are today. And so we'll need to figure out how to, to really deal with this, with this learning efficiency problem.

Harry Stebbings 3:50 You're probably thinking, "What did I get myself in for?"

Joelle Pineau 3:53 (laughs)

Harry Stebbings 3:53 Um, and I don't blame you. Uh, I ask questions that I think everyone else thinks, but I'm not afraid to say, "I don't know." Why is RL so inefficient?

Joelle Pineau 4:02 Um, there's a few reasons. You're going to get me on like a-

Harry Stebbings 4:06 This is great.

Joelle Pineau 4:06 ... a deep dive-

Harry Stebbings 4:07 I, I really love them.

Joelle Pineau 4:07 There's a f- there's a few reasons. One is the fact that RL is about sequential decision-making. So, you know, think about you're starting at a point, you need to figure out what you're going to do next, and you might pick the right side of the branch or the wrong side of the branch. And then, like, the road keeps on splitting. So, every time you make a mistake, it sort of compounds ...

Harry Stebbings 4:28 Mm.

Joelle Pineau 4:28 ... through the length of the series of actions you're making. So, that means, like, the amount of error you can make can be very, very large. And to get it right is quite difficult. Sometimes people compare it to, like, a needle in a haystack is like finding the right solution in RL. So, there's that part. The other part that's hard is the fact that to train the system, to train the models, you have to essentially take actions to learn. You can't learn from static data. You can learn some things from static data, but actually, you know, to, to get the right policy, you need to test it out. And so that means you need a simulator. You need to get the synthetic data. All of that can be really expensive also. And so we have an ... difficulty getting, like, just a variety of environments and simulation to test RL.

Harry Stebbings 5:16 When we look at the cost curve for RL, you said you've been working on it for 20 years.

Joelle Pineau 5:19 Yeah.

Harry Stebbings 5:19 Have we seen that dramatically come down? Will we see it continue to dramatically come down? Or is it a case of it is just a fundamentally expensive method of training?

Joelle Pineau 5:28 Hmm. Um, it's come down, especially in domains where we have good reward functions. So, the place where most people started hearing about RL is around the AlphaGo time. You know, the game of Go, which was sort of one of the goals for AI. Many people thought we were, at that time, we were still a decade away from being able to have machines play Go at the level of humans. And, and out comes a team from DeepMind, you know, uh, goes off, plays against, uh, (laughs) the world champion and shows that, that RL can basically do it. Um, and so ...I would say, you know, in cases where we clearly know what's the goal, we can write down precisely the reward function, we're good. We can make a ton of progress. So, that's why you're seeing progress in mathematics, um, very well-defined reasoning tasks, games, g- these kinds of things. RL to shape the behavior of models to get them to be social creatures, that we have no idea how to do. I mean, I don't know if you have children, but, like, shaping their behaviors, you know, the number of times you can repeat the same thing and still they do something else. And so, there's something there, you don't know how to write that out mathematically. And that's where I think we're still in for, for some hard work.

Harry Stebbings 6:46 (laughs) Okay. So, w- we still have some hard work. When we look at the training versus the inference market today, we've had so much weighed on training so far, and it's been incredibly costly and expensive. And then I hear everyone say, "Well, actually, inference is 95% of the market and that's where it's all going. That's where NVIDIA will make most of their money." How do we think about the cost curve applied to training versus inference and where it sits today?

Joelle Pineau 7:10 Hmm. I think there's a lot of different variance. And, you know, if you'll allow, maybe I'll pivot to, like, you know-

Harry Stebbings 7:18 Pivot anywhere.

Joelle Pineau 7:19 ... wh- where I'm, where I'm going with Cohere now.

Harry Stebbings 7:21 Yeah.

Joelle Pineau 7:21 Just, you know, I joined Cohere less than a month ago, super exciting company. I think one of the things that Cohere is doing is actually to develop AI models that run on premise, so that means enterprise bring it in, they run it locally. So, the company has to worry about the training of the models. Obviously, we want world-class models for the needs of enterprise. Doesn't have to worry about the inference. Doesn't have to worry about the inference cost. The, the, you know, the, the client's customers have to figure out what's the right way for them to digest the AI. That means, like, there's a lot of motivation to have very efficient models, uh, so that they can run really efficiently on premise. So, you know, we get caught up into one, one paradigm-

Harry Stebbings 8:05 Respectfully, if, if, if they're-

Joelle Pineau 8:05 ... but there are other paradigms as well.

Harry Stebbings 8:07 If they're the ones paying for the inference, is there not less incentive to make them efficient? Because sod it, you're not the one paying for it. If you're the one paying for the inference costs, God, I want it to be as efficient as possible because it's my dollar going to that. But if it's IBM's dollar, pff, I mean, yeah, I would love it to be efficient, but we're not paying for it.

Joelle Pineau 8:26 (laughs) We're still in the early days of AI adoption in enterprise, so what's good for the client is good for us. (laughs)

Harry Stebbings 8:32 Totally,

Harry Stebbings 8:33 totally get you. Um, what's the biggest challenge about capital-efficient AI today? I know that sounds strange, but when you look at the e- economics, so to speak, what's the biggest challenge?

Joelle Pineau 8:46 Um, there's a lot of challenge today. I think, uh, in terms of the economics of AI, I think one of the biggest challenge is the fact that it's very hard to have predictability, right? Everyone wants to know, when are we going to hit the breakthrough? Everyone wants to know, how many GPUs do I actually need? Everyone wants to know, like, what's the return I can expect? There's just a lot of uncertainty built into the system. A lot of that is because there's a lot we don't know about this technology. And so, that means we have to take in quite a bit of risk when you're building out. Whether you're building out your data center, whether you're building out your workforce, whether you're trying to figure out, you know, how much data to, to curate. And so, that makes it difficult for a lot of people. People want answers, and this is a world where we don't have that level of predictability compared to other industries.

Harry Stebbings 9:39 Does progression happen in kind of a linear fashion, or does it happen in step functions like AlphaGo, like, uh, Deep Seek, which, uh, d- depending, depending on kind of what you believe-

Joelle Pineau 9:52 Yeah.

Harry Stebbings 9:52 ... suggests a lot of efficiency in terms of model improvement. D- is it step function or is it linear?

Joelle Pineau 9:57 I tend to decompose different ingredients that lead to progress. You know, people often talk about, like, the algorithms, the data, the compute. I think, in general, compute and data have a more linear effect on progress. You build more compute, you run bigger models, you can typically get better performance, you feed in more data. It's not just quantity. You need to worry about quality and diversity as well. But roughly, it's more linear-ish with respect to the data. The algorithms are the ones that have the non-linear effect. And so, you can explore lots of ideas, and then something like the transformer comes along and just changes the paradigm. And it's not just your transformer, you know. On the optimization side, suddenly we hit upon Adam, which is a technique to, to do the optimization of your model, changes in the paradigm. Reasoning, suddenly, we start thinking about how to put that in-the-loop reasoning, and it changed the paradigm. So, those ideas tend to have a non-linear effect. The challenge with these algorithmic ideas though is that actually it may take a long time to prove themselves out. So, like, the paper can be sitting out there, there's thousands of papers coming out, the idea's sitting out there, and we may not think to try it with the right data, at the right scale, with the right combination of hyperparameters. And so, you don't notice that effect for a while. So, it's hard to predict, and it's non-linear more on the algorithmic side than I think on whether it's data, compute, even talent or other things.

Harry Stebbings 11:31 With respect to Google, I mean, Transformer's obviously birthed in Google and sat as papers for many-

Joelle Pineau 11:36 Mm-hmm.

Harry Stebbings 11:36 ... well, a couple of years.

Joelle Pineau 11:37 Mm-hmm.

Harry Stebbings 11:37 Um, when we loo- you mention there, you know, compute, algorithms, data, if we should just kind of go through them to understand, everyone suggests that it's weird there's two different worlds. It's like scaling laws exist, just throw more compute at it. When you look at data center investment, when you look at all the kind of desirability of compute, and then again you have GPT-5 seemingly focusing on efficiency and other signals, do scaling laws play out from here?... and if so, for how long?

Joelle Pineau 12:07 They've been... The scaling laws have been remarkably robust. Um, they don't play exactly as we expect, but still, they've been remarkably robust. Lots of people have bet against scaling laws in the past. And I would say overall, you know, we've seen a ver- pretty, pretty, pretty robust effect. Um, it... They don't work alone. We also need these algorithmic innovations. Um, but, but most of the time, you know, I w- I wouldn't bet against it.

Harry Stebbings 12:39 O- o- on the algorithm side, is that the hardest to innovate on? You could think about, "Well, you can buy more compute. It might be hard, but you can buy more compute and data. There are different ways, whether it's synthetic or human." Is algorithms the hardest to innovate on?

Joelle Pineau 12:53 It's certainly the most creative work to be done. And, you know, the space of ideas is so wide, um, that I would say, i- i- it's hardest in the s- sense that, like, you can move in... You know, I'm researcher at heart. You can move in so many different directions, and picking the right one, you don't know till you get there whether it was the right one or not. It's a little bit like reinforcement learning. (laughs) So, in that sense, I think it's, it's the, it's the most interesting one, it's the most frustrating one, and it's the most difficult one (laughs) , certainly from investors' point of view, to know where to put your chips.

Harry Stebbings 13:30 Is... M- Speaking of, kind of knowing where to put your chips, and moving from purely a research lens with, with Meta-

Joelle Pineau 13:36 Yeah.

Harry Stebbings 13:36 ... to now also building product-

Joelle Pineau 13:38 Yes.

Harry Stebbings 13:38 ... is there ever this, like, inherent conflict between intellectually interesting research with the need to productize and monetize, and how do you think about

Harry Stebbings 13:47 that?

Joelle Pineau 13:47 I mean, one of the reasons I w- I'm really excited to be joining Cohere, actually, is because, like, we're at a stage where AI is really starting to be useful. Maybe not as useful as people think it is, but we are there. Um, and by working on AI that's going into enterprise, I feel we're gonna get such an interesting signal of what works and what doesn't work. You know, we keep on talking about, you know, AGI and AI for the masses and so on. But, actually, like, when you need to sell AI to a business, you get a real signal of what works, what doesn't work, um, and that's what I'm most curious to see. Um, and, uh, you know, we've been using these academic benchmarks for many years, you get some signal. But it's not the same as, as getting this to do productive work. Um, so, I'm curious to, I'm curious to learn out of that. You know, we're gonna get new types of data. We're gonna get, I think, a lot of insights that are then going to drive the research ideas. Um, I think that's the, that's the other thing to think through. When you have a large space of ideas to explore, getting that feedback signal from the real world is super useful to guide you through that search of ideas.

Harry Stebbings 14:58 I just had a great chat to David Cohn at Sequoia, who said that he thinks a good barometer for utility value within enterprises is, like, does it have the ability to replace the work of your bottom 5% in any category?

Joelle Pineau 15:11 Mm-hmm.

Harry Stebbings 15:12 He says, like, "We overestimate a lot."

Joelle Pineau 15:13 Yeah.

Harry Stebbings 15:14 "Can it replace the bottom 5%-"

Joelle Pineau 15:16 Mm-hmm.

Harry Stebbings 15:16 "... in any function? And if it can, that's a very meaningful improvement." Do you think that's a good barometer, and how would you s- advise an en- enterprise on whether something's useful or not as a yardstick?

Joelle Pineau 15:28 Mm-hmm. I... Y- you know, I, I, I, I prefer, in terms of a barometer of, um, eh, productivity, something a little bit different, which is to say, can most of your employees do 10X the amount of work with AI versus on their own? That, to me, is actually a better barometer. I think human and AI have very complementary abilities, so to just, like, flat out replace a portion of your workforce is actually pretty unrealistic. Um, some may, some may try and some may be slowing down their hiring, but I actually think-

Harry Stebbings 16:02 Respectfully, I think 10Xing your work feels more unreal- is that, is that not a bigger ask? I'm almost more intimidated by 10Xing my work.

Joelle Pineau 16:09 Oh, I don't think that's unrealistic at all.

Harry Stebbings 16:12 Wow.

Joelle Pineau 16:13 Yeah.

Harry Stebbings 16:14 I- in a timeline that is next couple of years?

Joelle Pineau 16:17 Yes. Yeah.

Harry Stebbings 16:18 I'm sorry, how d- how does that actually shape out, then?

Joelle Pineau 16:21 I think you have to identify very concretely the types of work that you are delivering. But I think we're starting to see, like, you know, Hollywood-quality productions being made in a matter of hours. We're seeing, you know, to take a super concrete case, like machine translation. If humans are doing the machine translation compared to machines doing it, you'll go from hours to seconds on long-form text, multi-page documents. And so, for a lot of work, it's not like AI can do all of the work. Humans still need to ask the right question, they need to verify the information, they need to shape the tasks. But once the task is well-defined, the product are clear, like, all the design considerations are fed into the prompt, you press the button and you've got an answer in seconds for something that used to take sometimes weeks and months.

Harry Stebbings 17:14 So, completely hear you and understand that. I'm just kind of re- trying to re-evaluate a belief that I had for the last few months, which is, like, fo- I- I'm a venture investor first, and-

Joelle Pineau 17:25 Mm.

Harry Stebbings 17:25 ... for all of us to make money-

Joelle Pineau 17:26 Mm-hmm.

Harry Stebbings 17:26 ... we need to see the transition from kind of human labor budgets to AI spend.

Joelle Pineau 17:31 Mm-hmm.

Harry Stebbings 17:31 And it's with that transition where we obviously see the TAM massively increase-

Joelle Pineau 17:35 Yeah.

Harry Stebbings 17:36 ... and we make a lot of money. But when I hear you say that, I suddenly question that assumption as, like, the barometer for whether we make money, because you're suggesting that actually we don't replace the human labor budget, it just makes us 10X more efficient.

Joelle Pineau 17:49 Mm-hmm.

Harry Stebbings 17:49 Is that correct?

Joelle Pineau 17:50 Yes. And I think, you know, uh, eh, eh, people have s- l- you know, there's a lot of nuance to all of that. And some work will be harder to get that same level of efficiency gain, whereas other work, you'll see 100X in terms of efficiency gain. But I, I do think that for a lot of the work that's happening right now, that's absolutely feasible.

Harry Stebbings 18:10 Where do you think... Uh, I'm sorry, where is the efficiency gains most-... tangible?

Joelle Pineau 18:14 It goes back a little bit to this notion of, what are the tasks that we can specify well? So, in any case where we can be very precise about what a great result looks like, we'll be able to make that task automatic much more easily than tasks that are much more nuanced and, and have a lot of complexity.

Harry Stebbings 18:36 So it's ambiguity.

Joelle Pineau 18:38 Ambiguity in the specification of the task is what's hard for our machines.

Harry Stebbings 18:43 Mm. How have you seen enterprise reaction to this? There's fear from workers sometimes, there's excitement from leaders.

Joelle Pineau 18:53 Mm-hmm.

Harry Stebbings 18:53 There's apathy sometimes.

Joelle Pineau 18:55 Mm-hmm.

Harry Stebbings 18:55 How have you seen and measured enterprise response?

Joelle Pineau 19:00 A lot of the, the workforce can be reasonably fearful about job displacement. Um, there's a lot of also individuals who have, you know, a bit of an instinctive reaction to, to change. And, and change can be hard for a lot of people, and we're seeing a lot of change in a very short time span. And so, I think there's also a generational effect. I think for some generations, there's th- that, that change is, is more, more jarring. I think for the younger generations, I have teenagers, young adults at home, for them, AI is just native. They just, you know, kind of, you know, are gonna grow up with that technology in a different way than, than some of the older generations.

Harry Stebbings 19:41 It's interesting, Sam, uh, speaking of kind of children there at home and how they engage with it, you know, Sam Altman said that kind of young people engage with it as like an OS to the world-

Joelle Pineau 19:50 Mm-hmm.

Harry Stebbings 19:50 ... and AI is that companion for them. And then older people kind of use it as an next-gen Google.

Joelle Pineau 19:54 Mm.

Harry Stebbings 19:55 Do you agree with that, and do you see that in your work?

Joelle Pineau 19:58 Um, I see a lot of people using it as a tool more than as a, as a companion. Uh, you know, people have this, you know, Swiss knife in their, in their work life all of a sudden that can be super helpful, but that's really most of what I see.

Harry Stebbings 20:16 Totally get you. What are enterprises' biggest challenges with AI adoption at scale?

Joelle Pineau 20:21 For, for many enterprise, you know, one of the challenges is to make sure that the AI comes in and can be integrated in their workflows, their processes, their information. And so, the challenge is to deploy in a way that allows them to exploit all of the information systems that they already have. And some of them have accumulated these over decades. And so, that's, you know, some of the work that, that remains to be done, of figuring out how to do that.

Harry Stebbings 20:51 So it's integration with existing systems and data flows.

Joelle Pineau 20:53 I th- uh, and, and that's, you know, of course that's something we see a lot at Cohere because we do on-premise deployment. You know, one of the things we focus on the most is, um, data confidentiality and security so that, so that enterprise can exploit all of that information. So that's top of mind for us. Um, but it's also a huge opportunity, I would say. There's a, there's a big interest in that. But making sure to, to get that, that compatibility I think is a, is a challenge. Um, in many cases, for people, you just, you know, change is hardest for people, and so you have to get them curious about using the technology. Many people feel they have to get it right the first time, and, and I really think like a spirit of exploration and curiosity is, is much better suited to (laughs) to the phase of maturity of the technology that we have today. We don't have all the answers of how it should be used. That's gonna come from, from people on the field.

Harry Stebbings 21:51 Security is a topic that we quite often glaze over, especially when in- uh, investing in kind of application layer AI tools. What does no one know about AI security that people should know?

Joelle Pineau 22:04 With respect to AI security, I think there's a new front that's opening up with, um, the development of agents. And frankly, there's a lot we don't know yet in terms of the vulnerability of these systems. With LLMs, we're starting to get a better understanding. We've had, you know, quite a bit of red teaming exercise and jailbreaking and so on, and so people have identified different risk vectors, prompt injections, things like that, which are vectors for malicious actors to interfere with the system. With AI agents, we haven't seen that. And one of the features of computer security in general is, often, you know, i- i- it's a bit of a cat and mouse game, quite frankly. Like, there's a lot of ingenuity in terms of breaking into systems, and then you need a lot of ingenuity in terms of building defenses. And so, we just have to stay very active in that sense.

Harry Stebbings 23:00 What are the potential vulnerabilities though in an agent world?

Joelle Pineau 23:04 In terms of agents, you know, we, we worry a lot about, um, hallucinations in LLMs. The parallel in agents is impersonation, so agents that come along and are essentially impersonating entities which they don't legitimately represent. And in doing so, taking actions on the behalf of these entities where they don't legitimately represent, whether it's, uh, infiltrating, you know, banking systems and, and s- and so on. And so, I, I do think we have to be quite lucid about this, develop standards towards it, develop ways to, to test for that in a very rigorous way. There's ways to reduce that risk drastically. You run your agent, you know, completely cut off from the web, you're reducing your risk e- exposure significantly. Um, but then you lose access to some information. So depending on, depending on your use case, depending on what you actually need, there's different solutions that may be appropriate.

Harry Stebbings 24:04 Totally get you. That's a really hard one, because like then verification becomes the most important thing.

Joelle Pineau 24:10 Mm-hmm.

Harry Stebbings 24:10 But then it's like, who's the arbiter of verification? Is it governments? Is it, is it companies?

Joelle Pineau 24:16 Mm-hmm.

Harry Stebbings 24:16 How, how does one think about that?Who says you're a valid agent versus an invalid agent? (laughs)

Joelle Pineau 24:23 Um, governments can be good for defining standards on which we all agree. Companies are much better at building the solutions at scale and deploying them.

Harry Stebbings 24:36 Do you think governments are good at setting the standards? When you look at AI and where we're at, and then when you look at the sophistication levels of government, um, programs or decision-makers, with respect, they're just a little bit behind. Do you think they are actually equipped?

Joelle Pineau 24:55 I don't think you should look at where government are in terms of necessarily AI regulation. Like, AI as a field is so incredibly young and fast-moving, and, and by nature, and there's some good in this. Governments are moving a little bit more cautiously and, and, and in ... usually need to benefit from our knowledge to make good policies. Um, and so, I, I do think you, you can look at other fields in terms of, of regulation. You know, you look at, um, uh, aviation. The security record for aviation today compared to where we were 50 years ago is just incredible. And governments have played a role in defining that, in terms of standards and in terms of what are the, the norms and so on. So, I'm quite hopeful, uh, uh, I'm an optimist about this, maybe it's my Canadian side, but governments can play a useful role, um, in many cases. You know, clear standards actually means reducing uncertainty for a lot of companies in this space. But we shouldn't expect that to be ahead of the technology. I think that would be the wrong order of things. In some sense, we need to develop that technology with enough of a creative space, and we need to learn fast, and then develop the right, the right guardrails for that technology from, from, from the real earnings we have.

Harry Stebbings 26:12 Uh, we, we mentioned that kind of, uh, governments and their role. Uh, when I had Nick on the show, he was saying, actually, the benefits of not being an American company, um, given some geopolitical challenges sometimes. (laughs) Um, I'm just intrigued. Do you think we will have these government sovereign models for each geo? You know, we have Mistral in France, we have Cohere in Canada, founded in Canada, but then you've got global kind of HQs.

Joelle Pineau 26:36 Mm-hmm.

Harry Stebbings 26:37 Do you think we will have these sovereign models and regionalized winners?

Joelle Pineau 26:41 I do think it's healthy that there are models that are getting built in different places around the world, not just in, in the US and, and China right now. I think this is healthy in terms of diversity of, of thoughts. I think it's healthy in terms of having a greater amount of people with access to technology. I do think for Cohere, you know, the vision isn't to be a Canadian company. Like, the vision is to be a global AI company, and I think, yes, you know, we have a headquarter in, in Toronto. We have teams that are distributed around the world. We have a great team m- and here in London, as well as in the US, uh, and, and France and other places. And so, you know, having that ability to deploy models that operate across the world, I think, is gonna be absolutely an important part of the, the strategy for Cohere. I think there's a great opportunity. Um, a- what it gives us to be headquartered in Canada is like a s- sensitivity to the fact that it's not always a one-size-fits-all solution. You know, I go back to the research we've done, um, we've done leading work in terms of multilingual model. Um, and it turns out it matters. You go to Japan, you go to Korea, and they do want models that work well in their language. People in the workforce are still operating in, in the language of the country. So, having a company that, that is attuned to that, that values that, that internationalization of model is actually important on the

Joelle Pineau 28:06 global market.

Harry Stebbings 28:07 Totally get that. On the team building side of it, Canada has great talent. You mentioned, obviously, some in London as well. What have been your biggest lessons/observations on team building in this, like, talent frenzy that we're in also? How, how will you analyze that?

Joelle Pineau 28:23 One of the things that's, that's important when you're, you're building a, a team for AI, I do think you need people who have v- vision, who have, like, a sense of, like, "What can we create?" Just because we're in a space where there's so much innovation that is still needed. So, you need an ingredient of vision that can be one, two, three people who bring that, that ingredient of, of vision. You need people who have amazing execution muscle. Like, they don't care that it's their idea. They care that if the team agrees on an idea, they are just gonna push this and get it done. They're gonna build a system. They're gonna run the experiments. They just have that technical rigor to execute. And then you need people who kind of like keep the team together, who have, like, the sense of, like, who needs what to operate well and who are that social glue. You know, humans are still social beings, and that social glue in a team matters a lot. Where I've seen it fail is to have just sort of one type of person inside the team. Um, I don't think it, it becomes that productive to, to put a bunch of AI superstars all together in a room without the execution machine, without the social glue. I don't think you get necessarily the same results. So, I'm, I'm a big believer in building teams with, with diverse complementary, uh-

Harry Stebbings 29:46 So, you can't just buy the Galacticos?

Joelle Pineau 29:49 I don't think you need to. I think you, you really have to be thoughtful about putting people in a group. The other thing that helps a lot is for the team to have focus. You know, if it goes in all sorts of different direction, you'll lose the, that, that power that you get from, from people working together, so having a lot of clarity. What's the North Star? What's the goal? Where are we going? Even if over time that needs to change, but that level of clarity is required for everyone to be working in the same direction.

Harry Stebbings 30:19 Can I be so blunt? If, if-

Joelle Pineau 30:21 Please.

Harry Stebbings 30:21 ... you don't need to buy the Galacticos-... why, why do you have like an Andrew Tulloch, a Daniel Gross, an Alex Wang, and the Galacticos assembling? Like-

Joelle Pineau 30:32 Yeah.

Harry Stebbings 30:33 ... i- is, is that wrong?

Joelle Pineau 30:35 Y- you do, you do need a few of these, like, uber talents in the team. Th- there's a relatively, you know, short number of people who just understand this technology very deeply. You do need some of this talent, and if you can afford it, you should, you should get some of that talent. But you don't need all of your team. You n- you need, like, a, a, a team with complementary skill as well.

Harry Stebbings 31:00 D- does that create a good team? Like, if I gave you, you know, $10 billion to go build a team, and you could buy a couple of these luxury star players, I feel like it's Top Trumps cards for, like, sports teams, but you can buy a couple, does that create a good team when one is a $3 billion person, and then the rest are, are just average $50 million people?

Joelle Pineau 31:19 Yeah. Uh, I- I- I wouldn't say no if someone offers me the (laughs) opportunity to, to hire. There's definitely some really talented people in the field. And they deserve to be fairly compensated. This technology is going to have prof- make a lot of people very rich and have major effects in terms of society. And so, you know, we should be rewarding the, the talent. But I'd be very thoughtful about, what are the teams that I put together, and how do they work together, rather than just, like, you know, hire a roster of superstars without being thoughtful of how they're gonna work together.

Harry Stebbings 31:54 So, you, uh ... It's so funny. So, because of the impact that you can have in these teams, actually, the multi-billion dollar price tags that you see can even be justified.

Joelle Pineau 32:05 Time will tell. I don't think it's necessarily needed to go at that scale, but

Joelle Pineau 32:11 time will tell.

Harry Stebbings 32:11 If I gave you $10 billion, what would you spend it on first?

Joelle Pineau 32:16 One of the things you need is a balance between talent and compute. I think, you know, if you have too much talent and not enough compute, you're, you're, you're wasting your time. So, usually like an equilibrium between those, those two pieces. I think we often underestimate the importance of data. And, uh, data is getting more and more expensive. And so, I would certainly spend a good chunk of it on, uh, on data as well.

Harry Stebbings 32:45 So many things to unpack there. Um, do you feel you have sufficient compute today?

Joelle Pineau 32:50 I think we are reasonably well-resourced in terms of compute, uh, in, in, in building the models that we want to build. Yeah.

Harry Stebbings 32:57 So, access is not a massive problem?

Joelle Pineau 32:59 No.

Harry Stebbings 33:00 Okay, great. Why is data becoming more expensive?

Joelle Pineau 33:04 Data comes in, in different forms. And, um, on the one hand, you know, the, the notion, the days of, like, having data labelers who can say, "This is a cat," and, "This is a dog," are somewhat over. Like, the easy task, the AI can do. So, we're getting in a space where we need more specialized tasks. So, you know, imagine you're building AI for enterprise, there's a particular business logic. You need to make sure that you're catching the errors. You're gonna need someone with, like, deeper understanding of the tools. So, that's more expensive talent to come in and actually prepare the data. Um, there's also a lot of data that's synthetic data. So, when you're building agents, you need to build environments, and to build environments, you need some pretty creative folks who are going to build you, like, synthetic simulators. Um, we've seen this on the robot side for many years, b- people building robot simulators. Now, you're building AI for enterprise, so you need to think of, like, how are you gonna simulate these work processes in a reasonably realistic way that the AI can train on that? And so, that generation of environments and benchmarks and dynamic domains, uh, can, can be pretty expensive too.

Harry Stebbings 34:16 When you loo- ... Again, many things to unpack from that.

Joelle Pineau 34:19 (laughs)

Harry Stebbings 34:19 When you look at, um, the expanse of data, and then you said, "Oh, you know, cat, dog, uh, lamppost, we've got these capti- you know, captures?" You know, the, the, you click the ones which have like a z-

Joelle Pineau 34:30 Yeah.

Harry Stebbings 34:31 I get them wrong.

Joelle Pineau 34:32 (laughs)

Harry Stebbings 34:33 I legitimately get them wrong. I'm like, "Jesus-"

Joelle Pineau 34:35 They're getting harder.

Harry Stebbings 34:36 They're getting so hard.

Joelle Pineau 34:37 They are, they are. It's not just you.

Harry Stebbings 34:38 I, I, uh ... The other day, I called up my CFM, like, "I failed the Revolut, uh-"

Joelle Pineau 34:42 (laughs)

Harry Stebbings 34:42 "... I'm so sorry. I'll try again in a half an hour (laughs) ."

Joelle Pineau 34:45 (laughs) "Please let my A- AI agent answer that one for me."

Harry Stebbings 34:48 Honestly, it was embarrassing. Um, but, um, the question that I have is, you know, when you look at Core, when you look at Surge, when you look at Turing, how do you evaluate that market which is providing a lot of that talent?

Joelle Pineau 35:02 Mm-hmm.

Harry Stebbings 35:02 Is that an ongoing, enduring market? Or is that just a, "Hey, for the next three to five years, we'll need it in the training phase of these models, but I don't know what it looks like beyond that"?

Joelle Pineau 35:14 I don't think it's a phase in the sense that I do think this partnership, we'll call it, between humans and machines, where human provide guidance to machine, like, we are in this for a long time. What will change is the nature of the information that the AI provides versus the information that the humans must provide as a complement. And so, some of these firms may not be around in five years, but this notion of having humans guide the behavior, guide and train the behavior of the AI system, that is here to stay.

Harry Stebbings 35:48 It- it's super interesting. I kind of, um, as an investor in, in, um, one of them, um, see all of them converge around, like, needing to do three things now. They used to just kind of be talent acquisition. "Oh, we'll get you these people." And now they're like, "We'll get you these people, and we'll get you high-quality data that you can really use." And now, it's like, "Oh, shit, we need this third pillar," which is, "We'll also help you implement that data into your models, do training, and help you with benchmarking and proving that it's actually valuable."

Joelle Pineau 36:17 Mm-hmm.

Harry Stebbings 36:17 And now they need all three.

Joelle Pineau 36:19 Yeah.

Harry Stebbings 36:19 Are you seeing that third one where it's like implementation of their data as well, they don't just hand it over the fence?

Joelle Pineau 36:25 There's definitely some of that that's happening. I think, for me, the, the even bigger trend we're seeing is the moving from just labeling data to-... crafting environments, um, to-- to-- to-- to produce new-- new tasks.

Harry Stebbings 36:38 You said about synthetic data, and that also being a very important segment to consider. Do you d- get model degradation when you get this kind of reinforcing loop of models learning on synthetic data which sh- creates more data for synthetic ... And it actually degrades, or does it improve?

Joelle Pineau 36:56 It really depends how you're generating your synthetic data. So, in some domains, if you think like images, languages, like LLMs talking to each other at some point, you definitely get the degradation. And that degradation is due to essentially, like, a loss of diversity of your data. So, you know, you can make an analogy. Uh, you know, you'd, you take a bunch of people, put them on an island and let them reproduce. You know, at some point, the genetic diversity is gonna keep shrinking. Um, and so you get a reasonably similar phenomenon with, uh, with models because you're not injecting diversity into the data. So, for ... There are domains where lack of diversity means you get a collapse of distribution. There's other domains where, um, you don't need diversity. If you think of, like, you know, playing chess, playing Go, these kinds of games, we know exactly how to generate board configurations, and so we can generate tons of synthetic data. Not endless, 'cause it's a closed world, but still tons of synthetic data, and through that, learn for a long time. Then there's domains that are sort of in between. If I think of coding, we can generate synthetic code. You would take normal code. And we know how to inject diversity into the code. Like, I can take a couple of repositories, mix and match, apply an LLM to transform it. And so there's a way to generate synthetic data. The language is predictable enough, and there's enough structure that I also know how to inject diversity so that you don't get that collapse. So, the hope is that, especially in these domains, we can use a lot more synthetic data and do it without suffering from the degradation of performance.

Harry Stebbings 38:42 Do you worry that we are creating a world with just much worse code? Uh, a lot of people are concerned by the quality of code that's being outputted, and actually how we're just relying on it pretty haphazardly. Do you worry about that?

Joelle Pineau 38:55 Let me make an analogy in terms of like the quality of generation. You know, y- you ask about c- code generation, but let me take you back to 2015 and image generation. I don't know if you have it in your mind, but the quality of the images that were generated ... We had image generation models in 2015. They were really bad. The resolution was bad, the composition was bad, and so on. And from 27- ... 2015 to about 2022 or so, we saw huge progress in terms of the quality of the image generation. So, you think of code generation, like right now, we're in the phase we were for image ten years ago. Yes, there's a lot of bad code that's getting generated. Um, there's a lot of code that will get thrown away. But wait another ten years, and I think the quality of the code that's produced is gonna be excellent.

Harry Stebbings 39:48 Will we have d- ... What will the developer world look like in ten years when that is the case?

Joelle Pineau 39:53 (laughs) Well, if I carry my analogy further, I don't know if it's a reassuring scenario, 'cause if we look at where we are today in terms of i- image generation, there's just like ... The volume of image getting generated is huge. Uh, what matters now is sort of y- you know, picking the ... picking the quality out of the volume. And so, if I fast-forward ten years on code generation, when the ... we have the ability to generate a ton of code to do a ton of different things, we're going to need some selection mechanism to decide what code we actually want, when there's actually value. And so that's gonna come. There's still gonna be some sort of editorial, design choice. Someone needs to decide, like, of all the code we can generate, what's the code we want to generate? What do we need to be running in terms of our digital world?

Harry Stebbings 40:45 So, it's like a chief curation artist-

Joelle Pineau 40:48 Yes.

Harry Stebbings 40:48 ... within each company.

Joelle Pineau 40:49 Curation d- curation doesn't go away. G- c- curation, verification, this is work that doesn't go away.

Harry Stebbings 40:55 Does the structure of teams fundamentally change, then? Y- y- ... It's funny, kind of playing that back to you, and then also s- playing back to you what you said earlier about the human and g-

Joelle Pineau 41:03 Mm-hmm.

Harry Stebbings 41:03 If that is the case, there's not much of a partnership, is there, between human and AI? It's a chief curation person sitting on top of a huge amount of artificially created code.

Joelle Pineau 41:15 Well, that's your 10X productivity improvement there.

Harry Stebbings 41:18 It is.

Joelle Pineau 41:18 (laughs) It ... Uh, you tick in that box. It removes the human from the ... Um, you still, you still need people with, with, uh, with intent. That's one thing that, you know, you need to decide what you want to build and what purpose does it serve. And so that, that intent is still there. That, that role of, of critique is still there. Um, so the, the team composition does change significantly once you suddenly have, you know, designers who, in their hand, have amazing tools to dif- ... Go directly from the ideas in their head to the, to the digital world, maybe eventually to the physical world. That equation definitely changes.

Harry Stebbings 42:01 Do you think prompts and the way that we interact today with prompts from a chat largely, is like the enduring interface for human engagement with AI?

Joelle Pineau 42:12 It's awfully limited. And, you know, p- prompts can mean a few different things, but, but the idea of like typing in a box, that to me is very limited. And we're gonna break out of that box. Already, we're seeing a lot of, uh, cases where voice is a lot more natural as an interface. I do expect we'll see, you know, gesture, eye gaze, these kinds of much more multimodal ways to interact with, uh, with the AI rather than just stick in that, in- in- in that prompt box. But language is incredibly powerful. So, if you think of ...... prompt as being more language, as a way to express ideas and communicate with a machine, that's a powerful paradigm. I mean, as humans we ... So much of our communication is based on language, I don't think we're going to move away from that, uh, because it encodes information. You know, language, words are symbols that encode so much information so efficiently. And so, I don't think we're, we're close to getting away from that.

Harry Stebbings 43:13 It's funny, this conversation has changed a lot of, kind of previously held assumptions for me. When you think about what you did believe, that you now have changed your mind on, what's most prescient?

Joelle Pineau 43:25 I'm a scientist that is happy to be proven wrong any time, as long as there's new evidence. I'm genuinely curious to know. Other scientists are much more, like holding on to very, very strong conviction. I have weak conviction, but very strong respect for the scientific method and, and, and rigor. Experimental rigor, you know, theoretical rigor as well. Um, so there's a ton of things. I mean, I ... Ah, f- I mean (laughs) ... I used to be quite skeptical that neural networks were necessarily the ultimate solution to machine learning. I'd seen enough cycles of neural networks kind of peaking and, and, and then, um, being less, less useful, and I used to think every time you change the scale of the data, you know, you go from hundreds of examples to thousand, thousands, to hundreds of thousands, to millions of examples. Every time you change the size paradigm that neural networks were the first thing we tried, 'cause they're a universal function approximator, and then something else comes out that was better, and that was true for the previous generations. You know, some of ... You may remember SVMs as like-

Harry Stebbings 44:31 Yeah.

Joelle Pineau 44:31 ... being better than the neural networks in, in early 2000s, and I, I seem to be quite wrong on this one. Like, uh, neural nets seem to be here to stay (laughs) , um, and the ability to do back propagation and gradient descent and all that seems to be a really powerful way to learn.

Harry Stebbings 44:47 What does everyone else believe quite strongly that you think they are quite wrong on?

Joelle Pineau 44:52 I don't have a lot of patience as a scientist for people who are predicting sort of the, the extremist scenarios. Whether it's, uh, the catastrophic risks of AI or whether it's the, you know, winner takes all, you know, AI becomes our overlord kind of scenario. I don't have a lot of patience for that, um, and I wouldn't say it's necessarily widespread, but I just ... I think, uh, you lack scientific rigor to, to analyze these kinds of scenario. I'm, I'm much more pragmatic, grounded. I'm pro-innovation, I'm excited to see where, where AI is going and the problems it can solve, um, but I'm not so interested in just going around and, you know, making up science fiction scenarios.

Harry Stebbings 45:42 You've been on the most incredible ... You said there about kind of image generation in 2015 and, uh, funny how much it's improved. Um, we're seeing this kind of unbelievable capital supply go into-

Joelle Pineau 45:54 Mm-hmm.

Harry Stebbings 45:54 ... the space.

Joelle Pineau 45:55 Yeah.

Harry Stebbings 45:55 In a way that we haven't seen obviously for many, many years. Is it a good bubble, where we are getting incredible improvements and it's fundamentally advancing technology? Or is it a bad bubble, where costs are becoming too exorbitant, teams are too impossible to build, compute is too difficult? Is it a good bubble or a bad bubble?

Joelle Pineau 46:14 I think about it as a bubble with bigger variance. It's like ... (laughs) You know, the, uh, the upswing is gonna be bigger and, you know, there's going to be big downswings as well, and so there's a lot of variance into the system right now, um, as long as people have a tolerance to risk, then I think AI is a great investment and, you know, we should continue to be supporting, you know, risk-taking, new enterprise, new ideas. There's a ton of exciting new startups being created, we should continue to, to support them. You just have to be tolerant to risk.

Harry Stebbings 46:51 I've had some people on the show suggest that, uh, evals are, to put it delicately, bullshit-

Joelle Pineau 46:57 (laughs)

Harry Stebbings 46:57 ... uh, and that they don't actually mean anything anymore, and like, you know, humanity's last test, like, what does that really even mean? Uh, and we have these new tests that come out of it, what is this? And leader boards, what is this? Is that fair? Or do you think they actually serve a very effective utility to the ecosystem?

Joelle Pineau 47:15 I do think they, they are, um, really good indicators. So, I think you do need to take evaluations seriously, in terms of knowledge, but you shouldn't take them seriously in terms of the ultimate goals. So, you know, uh, evaluation, and there's lots of different benchmarks and so on, you have to decide like what type of models or model are you building? What's the characteristics of your system? And then think of evaluations as like unit tests for the performance of your system. I mean, software engineers will (laughs) will know what that is, right? Like, you run, sort of that evaluation and that gives you like a signal of how the system is doing in a particular dimension. But as we're building systems that are more and more general, do very specific tasks, you don't optimize for these, right? Like, I mean, we build AI systems that go into enterprise. None of our clients ask about like, "Are you able to win the Math Olympiad with this model?" That's not what they care about. They care about bringing value to their business. Now, we're curious to know how well we do on math problems because it can be predictive of behavior on other things, but you don't obsess over specific benchmarks. You kind of look at the ROI in terms of what you're trying to build.

Harry Stebbings 48:28 Can I ask, we mentioned there about kind of access for enterprises. Enterprises have money and that's a great luxury in a lot of cases. R- research institutes, universities often don't. With the kind of bubble like tendencies, people with money are able to afford to compute the talent. Are we seeing this kind of-... lack of access or democratization for, uh, great institutions that are educational maybe, who now can't afford to compete in this new world.

Joelle Pineau 48:54 Certainly, a lot of universities have a lot less resources than, than companies today. That's not completely new. Um, when I joined Meta in 2017, um, one of the reasons I did that is because I, I could already see the, the disparity in terms of access to compute. And I was really curious to see how you could do research with a lot more, uh, compute. But, you know, there's still amazing research that's being done in, in universities. You go to the major international conferences, NeurIPS, ICML, and others, and often the best paper awards are actually won by researchers out of universities. There's a lot of good ideas that you need to test out at small scale. Um, and in a university, you have a lot more freedom to pick pretty risky ideas at a small scale. But still, you know, no one's asking you (laughs) to, to, uh, to, to justify your, your, your research in, in ways that, that often happens in companies. So, I think they have ... They play different roles in the ecosystem, and what's actually specially good is talent flows between, between them. You know, university students come in, do internships, take jobs at companies. Uh, we've also seen a movement of people coming out of these large companies, going back to university, teaching, sharing with the next generation what they've learned.

Harry Stebbings 50:12 How important is it to have seen success and how valuable that makes you? When you look at people like, uh, Amira raising 2 billion out of 10 billion, it's like, "Well, no one's seen the success that she's seen with OpenAI, so it's valid." H- help me out as an investor. Is it that valid to pre- place that much of a premium on access to seeing it at that level that we are? Or is that slightly overpricing it?

Joelle Pineau 50:41 I think in many cases, you know, when it comes to, to, to d- deciding wh- where to invest very early on when you don't have tangible, tangible information, I mean, you look at people's track record. And there's a part of that that may be like, you know, what have they learned in terms of the, the core recipe? But the other thing is also the achievement of having put together amazing teams who are building world-class models, and there's a lot of subtlety to that. Um, and so I think, you know, both of these, um ... Both of these ingredients are, are important to consider.

Harry Stebbings 51:16 If you were investing today and you were joining my team, which category would you most like to invest in? Be it security, be it, um, generative AI, be it, um, compliance. You, you name it.

Joelle Pineau 51:30 Yeah. There's a lot of verticals, whether healthcare, scientific discovery, that I think have incredible promise where we're gonna see real, tangible progress within five years that are going to change completely the face of what we can do. So that's probably where I'd, where I'd push.

Harry Stebbings 51:45 That's very exciting on the healthcare front in particular.

Joelle Pineau 51:47 Mm.

Harry Stebbings 51:48 When you think about that timeline as well.

Harry Stebbings 51:50 Um, I'd love to do a quick fire round with you, if that's okay.

Joelle Pineau 51:53 Yeah, sure.

Harry Stebbings 51:53 So I'll say a short statement. What would you most like to do, but because of technical or financial limitations, you're not able to?

Joelle Pineau 52:01 I'm super keen to figure out how we build, um, societies of AI agents. We're doing it implicitly, but how do we look at populations of AI agents interacting together and having, like, a sandbox for, for doing that? Um, so, maybe something I'll, I'll get to do. Is it lack of time, resources, something else? There's just, like, a ton of different things to do. But keen to, to see what happens there.

Harry Stebbings 52:27 When you think about that ecosystem of agents, uh, you have children.

Joelle Pineau 52:32 Yes. (laughs)

Harry Stebbings 52:33 And AI changes our relationship with other humans-

Joelle Pineau 52:36 Yes.

Harry Stebbings 52:36 ... and friendship and social.

Joelle Pineau 52:38 Yeah.

Harry Stebbings 52:38 How does AI impact social friendship connection, do you think?

Joelle Pineau 52:43 Hmm. It definitely does, and you know, there, there's a sense that we spend a lot of our time in the digital world. And for some folks, you know, I look, two of my children, they spend a lot of time in the digital world, playing online games with their friends. It's still very social. It's ... There must be some AI, there's the digital platform, but it's still a very social experience. Others have more individual experience. There's definitely a shift of this time we spent, uh, towards that platform where we go look for that, for that social element.

Harry Stebbings 53:13 Knowing what you know, what do you not let your children do?

Joelle Pineau 53:17 Ah. (laughs) Eat too much sugar. (laughs)

Harry Stebbings 53:21 Sh- to- totally. That's, so that's the, like, physical diet.

Joelle Pineau 53:25 (laughs)

Harry Stebbings 53:25 Completely agree with that.

Joelle Pineau 53:26 Yeah.

Harry Stebbings 53:27 Is there a technical diet? I don't-

Joelle Pineau 53:29 Um, uh, I spend some time discussing, like, settings. I mean, like, you get an Instagram account, great. You can have an Instagram account. But, like, what are the settings on that account? Making sure they understand. Uh, I mean, they'll go and change them if they want.

Harry Stebbings 53:43 That's a fun conversation with mum, isn't it, "We're gonna discuss settings." Oh, God.

Joelle Pineau 53:46 (laughs) Uh, I know. That was not a popular one.

Harry Stebbings 53:49 Do they listen?

Joelle Pineau 53:51 Um, the thing with children is you don't know till later.

Harry Stebbings 53:56 Do you limit screen time?

Joelle Pineau 53:58 Um, I spent a lot of energy, especially in their younger years, limiting screen time. S- my kids did not have a phone till they were 14, 15.

Harry Stebbings 54:08 Did you see Adolescence?

Joelle Pineau 54:11 I have not.

Harry Stebbings 54:12 Okay. Watch it. It's, it's fascinating.

Joelle Pineau 54:15 Yeah, yeah.

Harry Stebbings 54:15 Uh, basically, a little boy goes up to his bedroom and, um-

Joelle Pineau 54:18 Yeah.

Harry Stebbings 54:18 ... gets lost down, kind of, rabbit holes of TikTok-

Joelle Pineau 54:20 Yeah.

Harry Stebbings 54:20 ... and Reddit, and-

Joelle Pineau 54:22 Yeah, I've-

Harry Stebbings 54:22 ... it does not turn out well.

Joelle Pineau 54:23 I have, I've heard about it. I just haven't had time to, to sit down and watch it.

Harry Stebbings 54:26 Do you worry about the, the loneliness pandemic, and then also just the mental health crisis that we have?

Joelle Pineau 54:31 I do worry a lot in general about making sure that, that people are healthy mentally. Yes, I, I d- I mean, I, I think w- the, the ... We have to be careful about taking shortcuts and saying, you know, because suddenly, you know, we have certain platforms, we have AI and so on, that is causing that mental illness. You know, there are a number of people who are suffering, and, uh, they deserve to-... to have g-good answers to, to, to the situation and they deserve to f- we deserve to find real solutions to that. And there's a lot of people looking for shortcuts and short answers, but I think more research into that is definitely warranted.

Harry Stebbings 55:12 What's your biggest lesson from working with Zack?

Joelle Pineau 55:15 He is incredibly deep into understanding the work. Like, he does not coast, you know. Uh, eh, when he started getting into AI, just the, the depth of the question that he'd ask, he just gets really interested in the topic and goes super deep, and that then just informs everything he does after. So, you can have the most amazing team, but like, as a leader, you need to go deep and understand the work.

Harry Stebbings 55:43 Did you see him change?

Joelle Pineau 55:45 As anyone gets more knowledgeable about a topic, you get more decisive. There's a phase where you're really learning and trying to, to understand, and there's a phase where you understand a lot of things and then you make your decisions faster. So, certainly that, that shift happened.

Harry Stebbings 55:59 What one AI buzzword would you ban if you had a magic wand?

Joelle Pineau 56:04 Existential risk.

Harry Stebbings 56:07 (laughs) Why?

Joelle Pineau 56:08 Because it just makes people afraid. And it's not out of fear that we make our best work and we take good decisions.

Harry Stebbings 56:16 Do you find the cost of talent is exorbitant?

Joelle Pineau 56:18 Talent is costly. Talented people deserve to be, to be paid well. And again, you know, uh, someone coming in just because of money rarely is going to be the right person, but you do need to compensate people fairly.

Harry Stebbings 56:30 Totally get that. Final one, what are you most excited for? Y- You don't like the existential risk. I don't like, like, the doomsday planning. When you think about the positivity that can come-

Joelle Pineau 56:41 Yeah.

Harry Stebbings 56:41 ... what are you most excited for when you look forward to the next three to five years?

Joelle Pineau 56:45 I do think s- some of the work in terms of, um, AI for, um, a scientific discovery is going to be pretty fascinated, pretty fascinating to see. Um, just in terms of the doors it's gonna open up, the ability to explore, you know, the combinatorial space of, of solutions. So, I'm curious about that. Um, and then I'm super curious to see, you know, how can we actually make our models more efficient? There's larger and larger and larger models. No one wants to run these models, you know. I spent a lot of my career building, um, open source models, and I'll give you one example. You know, w- we were in, in the frenzy of large language models and I pulled the stats on, you know, most downloaded models of last month. We had a model like Roberta from 2019, small language model was getting 20 million downloads a month. People want efficient models that they can use, that they can run, so I'm also super keen to see what we're going to be able to do at the scale that runs on like one or two GPUs.

Harry Stebbings 57:55 Final, final one. Y- you said there about kind of open, um... We seem to like be reverting to a closed world now. Is that the world with which we should predict and plan on?

Joelle Pineau 58:06 That's a mistake. That's a deep mistake. I mean, I will continue to believe that, specially for research, the ideas need to circulate, and this thought that you can just like close this down is, is absolutely false. I mean, people are circulated-

Harry Stebbings 58:21 Do you not think we are, do you not think we are though moving into that world? Like, everyone seems to be closing systems, closing access, and...

Joelle Pineau 58:27 There are definitely a number of places, people, that are closing down access. I don't think that is going to, to be effective. Ideas will circulate, and I also think it's a mistake from a point of view of fostering innovation.

Harry Stebbings 58:43 This has been such a joy. I've learned so much from this conversation. Thank you so much for putting up with my very basic questions.

Joelle Pineau 58:49 Yeah.

Harry Stebbings 58:49 But I've loved having you on the show.

Joelle Pineau 58:51 My pleasure. Thank you.