Joelle Pineau on Why Sovereign AI Just Got Real

00:00:11:08 - 00:00:30:19
Unknown
Hello, everybody, and welcome to AI inside the show, where we take a look at the AI that is layered throughout the world of technology. I'm one of your host, Jason Howell. Jeff Jarvis, unfortunately, is out today. So, you know, I'm sitting here in my studio flying solo, but I've got an amazing interview for you.

00:00:30:20 - 00:00:59:08
Unknown
My guest is Joel Pineau. For years she ran fair, which we've talked about. Jeff's definitely brought up fair a lot on the show. Meta's fundamental AI research lab, one of the most respected research operations in the field. Last year, she walked away from that to become the first ever chief AI officer at cohere, which is a Canadian company that's making a very different bet than a lot of the big US labs out there.

00:00:59:10 - 00:01:34:17
Unknown
They're not, you know, chasing AGI. They're developing enterprise AI for efficiency, security and in many cases, openness. She's a lifelong champion of open science. Her career actually started in health care, AI, intelligent wheelchairs and seizure detection, which we actually discuss and tie into her current role at cohere. She's also very outspoken about the need for sovereign AI, and, well, the timing could not be any crazier just a few days ago.

00:01:34:18 - 00:02:01:19
Unknown
You've probably already heard the US government forced anthropic to cut off foreign access to its two most popular models. Anthropic pulled them entirely for everyone. And mind you, this is all at the point of this recording. News may have changed because this isn't going to go up immediately, but this is a real argument for sovereign AI. Joel shares her thoughts on that and how it actually impacts companies like cohere.

00:02:01:19 - 00:02:28:19
Unknown
And cohere has been on a tear as well. Open sourcing powerful coding model. Just last week, a move into drug discovery deals putting its agents inside banks. Oh, and Joel picked up an honorary doctorate and gave a commencement speech about facing uncertainty in the industry, which is where we're going to start. So let's dive right into it. Here's my interview with Joel Pinot.

00:02:28:21 - 00:02:31:19
Unknown
Joel, thank you so much for hopping on AI inside today.

00:02:31:19 - 00:02:36:14
Unknown
It is such a pleasure to meet you and have you on the show.

00:02:36:16 - 00:03:15:24
Unknown
Yeah. And this is this is kind of a well, you've had a big week. I was kind of looking through your LinkedIn, of course, and you just posted. First and foremost, congratulations on receiving the honorary doctorate from the University of Ottawa just last week. That's pretty. That's amazing. Congrats.

00:03:15:26 - 00:03:38:27
Unknown
And to be in a position to be able to kind of share some of your knowledge and inspirational, you know, your wisdom and everything. It's a it's a cool opportunity to do that. You shared kind of your your entire speech in text on your LinkedIn. And I was reading through it. And, you know, the kind of focus of it is optimism in the face of uncertainty.

00:03:39:02 - 00:04:14:18
Unknown
You said that the graduates are stepping into one of the most uncertain and consequential moments in human history. Certainly something we've talked about. Me and my co-host Jeff Jarvis, who could not be here today, he would would have loved to talk with you, but he had something come up. But we've definitely talked about that uncertainty, and I have brought up a number of times on this show about what it's like to be a graduate in this moment, like never in my lifetime has it been the case where you go to school, you work super hard to to make all of these achievements and set yourself up for what your future, what you think your future

00:04:14:18 - 00:05:50:14
Unknown
is going to be. But here is this like moment in technology that really kind of puts a big question mark on things, whether rightfully deserved or not. I'm curious to hear your perspective on that.

00:05:50:16 - 00:06:12:01
Unknown
Have you? I mean, being that you were there and with them and talking with them and I'm sure, you know, have shared experience with them, like is is there anxiety for the folks in these positions, or is it excitement because, you know, with with a lot of change comes a lot of unknown directions? It's not it doesn't all have to be negative or bad.

00:06:12:02 - 00:07:09:22
Unknown
I don't know what what was the what was the tone, let's say.

00:07:09:25 - 00:07:29:28
Unknown
Yeah. At a time when the tools and how we do things are shifting so drastically and, you know, you see on the, on the side of schools, you see the one approach which is, you know, absolutely not. The way we teach and the way you learn is a certain way. And so these tools are, you know, not allowed or not welcome.

00:07:30:01 - 00:08:44:28
Unknown
You see a whole other kind of approach, which is, well, but if this is the way of the future, we're doing you a disservice by not teaching you like healthy skills or responsible ways to use these tools, because that's really that really seems to be the direction that this is all headed, and that you kind of need to know those skills.

00:08:45:01 - 00:09:06:27
Unknown
Yeah, 100%. It's not just the the students trying to figure their way through the tools. I was just yeah, talking with, you know, at a much different scale, but talking to like a high school teacher friend of mine and just kind of the challenges of, of, on one hand, recognizing the value of these things and being like, this is amazing technology.

00:09:06:28 - 00:09:28:28
Unknown
It's incredible what we can do with it and what, what is possible. But on the other hand, just not having the built up kind of facility around how how this like transitionary period between before and after, how this happens. And it's almost like I want to I want to skip forward like five years or ten years down the line and see where things where things land.

00:09:28:28 - 00:10:33:22
Unknown
Where do you think things are going to land like in that regard, based on where we're at right now and the direction that things are headed?

00:10:33:25 - 00:10:53:25
Unknown
Yeah. And along those lines, you you told grads kind of similar to what you were saying, say yes to the role that they're not quite ready for that. That's that speaks to me because I've done a lot of, you know, I used to work for a couple of companies, you know, bigger companies doing podcasts and stuff for them.

00:10:53:25 - 00:11:14:14
Unknown
Only a couple of years ago when independent and faced the kind of difficult and uncomfortable, you know, scenario of saying yes to things that I'm like, am I going to be able to do that? Well, I'm just going to say yes to figure it out. So there's a lot to be learned there. So that's very valuable and valid advice.

00:11:14:14 - 00:11:44:16
Unknown
But, you know, at cohere and elsewhere, building agents that are changing what things like entry level roles actually are, you know, North Mini code that launched just last week, which congratulations for that. That's awesome. We'll definitely talk about that. But I'm I'm curious like how much of what a like a junior engineer does can already be handled by a tool like North Mini code today.

00:11:44:16 - 00:13:32:21
Unknown
And where does that leave like the risk and the the optimism under uncertainty equation. Like where does that risk go. Is that risk falling on the hands of the The graduate, let's say. Or does that fall on the company that's building the tool who shoulders more of that risk I guess, is what I'm asking.

00:13:32:24 - 00:13:59:18
Unknown
Yeah, I've, I've definitely on a, on a personal level that myself to kind of gone through the general arc of like, oh my goodness this is amazing. I can just yeah, you do all that stuff and I'm going to be over here that go check in, I'm going to take that, I'm going to going to use it. But removing myself so far out of the equation that like there's no knowledge there, there's no way, you know, on ramp for me as the human to kind of pick back up with the with the detail there.

00:13:59:18 - 00:14:17:22
Unknown
So keeping the human in the loop, like you said, super important. Maybe, maybe that's a, I don't know, maybe that's a transition or a transformation that everybody kind of has to go through as they learn how to work with these models, because it's really tempting to be pulled deep down the, okay, I trust you to do everything. You got it.

00:14:17:24 - 00:15:12:18
Unknown
It's cool. Yeah.

00:15:12:20 - 00:15:48:26
Unknown
Now circling back a little bit to the optimism under uncertainty piece, as you can tell that before the first break. Like I'm really kind of dialed into your speech there because a lot of it really, really spoke to me. And I, and I look at kind of where, where a lot of this development is happening right now and this this idea of being optimistic, even through uncertainty, has the has the possible temptation of leading people to really over explore or never quite, you know, they just keep spending and spending like I see this a lot in the news.

00:15:48:26 - 00:16:07:06
Unknown
Keep spending trying new things. There's a lot of that happening in the industry right now. And I think looking at the state of the industry, everyone seems to be chasing a lot of the same bets they're making the, you know, the same assumptions and over exploring in that regard. Do you do you think that the field is over exploring right now?

00:16:07:06 - 00:17:22:19
Unknown
Is everyone kind of going in the same direction or like what's your thoughts there?

00:17:22:21 - 00:17:46:07
Unknown
Yeah, well, and you've talked a lot about about like algorithmic monoculture. You've talked about diversity in these in these models and, and you know, this kind of plays into that. Right. Like where is the actual diversity happening right now? It seems like a lot of a lot of companies, a lot of people are focusing a lot of their time on the LM side of things.

00:17:46:07 - 00:20:45:26
Unknown
You've got folks like Yan LeCun focusing on world models and exploring it from a different perspective. So there's a little bit of diversity going on, like what are your thoughts there? Like what? What is truly diverse in the exploration of AI that's happening right now? Like what's what's kind of catching you?

00:20:45:28 - 00:21:03:10
Unknown
Love it. All right. We're going to take a super quick break, come back and talk about, you know, you talked a little bit about kind of your background, you know moving into cohere. So we're going to talk a little bit about where you came from, where you are now and what cohere is up to. So that's coming up here in a moment.

00:21:03:12 - 00:21:24:05
Unknown
Hey, I hope you're enjoying my chat with Joel Pineau, but really quick, I want to thank you. You know. Yes. If you're watching and listening, I want to thank you. But I also really want to thank the patrons because without the Patreon, the show would not be a show. Like, it literally helps support the health of this show.

00:21:24:05 - 00:21:42:23
Unknown
So I want to throw out a huge thank you to Peter and to Casey Kamiyama. Thank you. Just a couple of our amazing supporters on Patreon who help us do this show each and every week. You know, they get access to all the extra perks that we have, including a daily podcast called AI Inside Daily. Yes, that's right.

00:21:42:23 - 00:22:03:06
Unknown
If you support on Patreon, you get access to five days a week of new news related episodes on top of the regular episodes that you watch every Wednesday and some of these interview episodes. So it's a lot of episodes. Thank you in advance for supporting. We couldn't do the show without you. And yeah, so patrons, you're awesome. Thank you so much.

00:22:03:07 - 00:22:08:19
Unknown
All right. Going to take a quick break then come back with more with Joel.

00:22:08:21 - 00:22:34:22
Unknown
All right. Joel, you you left one of the, you know, a very prestigious lab at meta, the fundamental AI research team we've talked about on the show many times. Fair to join cohere just last year. And that was I don't know, that was that was a big bet on the highest upside that that we're tracking right now. What what made this transition worth betting on for you?

00:22:34:24 - 00:24:44:15
Unknown
This was a big transition for you.

00:24:44:18 - 00:25:53:15
Unknown
So on the flip side of that, what did you have to like? Was there anything that you had to give up like that that I don't know that that comes to mind in that transition? Something that I don't know was a little challenging to to give up as you transitioned over.

00:25:53:18 - 00:26:23:14
Unknown
Yeah. Yeah. Indeed. Indeed. Now fair was, you know, built around this idea of publishing everything very, like you said, open research, culture, that sort of stuff cohere, you know, definitely a different kind of focus selling enterprise security, enterprise control, that sort of stuff is was there any sort of. Well, I don't know. When you when we're talking about the open science part, was there any trading away or, or bringing that over with you?

00:26:23:15 - 00:27:08:07
Unknown
Like what does that look like?

00:27:08:09 - 00:27:28:21
Unknown
Amazing. Good. Well, then you landed at the right place. We talked a little bit about North Mini code, and maybe we can, you know, while we're talking about kind of like openness, maybe we can come back to this a little bit. A lot of us labs right now are kind of in a moment where they're tightening access around it.

00:27:28:21 - 00:29:55:09
Unknown
And, you know, on the flip side, you put the weights that they're in the open. A lot of what cohere sales to enterprises is, you know, in some ways can be the opposite of that security control, keeping things in-house. How do you decide what's worth opening up and then what kind of stays closed? What does that decision process look like for your team?

00:29:55:12 - 00:31:39:27
Unknown
Yeah, 100%. That's definitely the beauty and the value of of openness and open source and and all that stuff kind of hitting touching back to your speech a little bit, you said that there's a risk in sharing ideas openly or there can be a risk anyways, but that the upside is actually worth it. So when Cahir opens, chooses to open a model or a part of a model, what what is the risk that you're accepting in that, in that scenario?

00:31:40:00 - 00:32:41:10
Unknown
Is there a is there a particular downside to opening in that regard or in that situation. Probably dependent. Yeah. It depends. There could be.

00:32:41:13 - 00:33:06:06
Unknown
It's occurring to me as I as I'm listening to you talk about this, it it hadn't really crossed my mind until now. The amount of model development that I mean, I mean, clearly, you know, companies like cohere are developing lots of stuff that we never even get the light of data on as just general people. But it's never occurred to me like what those models actually are, like the thing that was created that the company decided.

00:33:06:07 - 00:33:24:25
Unknown
And you know what? We're just not we're going to put that out there for whatever reason. It's very intriguing. I mean, there's probably some really interesting, interesting details that you probably can't even talk about. But yeah. Two. Good. Yeah.

00:33:24:27 - 00:33:43:12
Unknown
Yeah, indeed.

00:33:43:14 - 00:34:27:28
Unknown
And maybe the idea the destination is like understandable and makes sense. But the this particular version of the destination isn't the right version to go. So maybe we tackle it from a different perspective. It's just. Yeah.

00:34:28:01 - 00:34:52:06
Unknown
Yeah. No, that makes sense. Shifting gears a little bit, 25 years ago you were building. A while ago you were building an intelligent wheelchair. You had a robotic nurse. You want a Governor General's Award for AI and personalized medicine going back in time a little bit flashing forward to last month and go here. Bought biopharma company reliant AI.

00:34:52:08 - 00:37:17:12
Unknown
Folded it into North for pharma. And I don't know that's that's kind of cool. It's kind of like the you know, the the back, the back catalog, the back history kind of meeting the new moment that you were at. How much of that was a coincidence, how much of that was something that you were directly and and deeply involved in?

00:37:17:14 - 00:37:45:01
Unknown
Yeah, that's super cool. I mean, after 20 years in reinforcement learning, kind of tracking with each other. Now, your colleagues, that's got to be really nice. Yeah. What of all the places that like. And Jeff and I talk about this on the show all the time, I firmly believe there's a lot of things that that AI, the way we're we're seeing it developed and kind of its strengths and everything can be incredibly applicable to.

00:37:45:02 - 00:38:07:13
Unknown
And there's other ways it's like, okay, yeah, sure, it can do that. Whatever. When it comes to medicine. Like that's that's just one, you know, medicine and health. And just like that seems like a technology that is made for that. Cause it's also though, a place where a wrong answer can, can have the highest cost, let's say, is the field.

00:38:07:14 - 00:40:21:20
Unknown
Do you think the field is ready to put agents, agents, AI into this, this aspect of things in a drug discovery and a clinical work, like, are we at a point to where agents in medicine makes sense, or do we still have a ways to go there?

00:40:21:22 - 00:40:25:22
Unknown
Optimizing?

00:40:25:25 - 00:40:59:15
Unknown
Whether it's.

00:40:59:18 - 00:41:21:10
Unknown
Yeah, so much development happening, what do you like kind of touching back on, you know, the 20 years ago versus now? What can I do now in medicine that you wouldn't have even been able to I mean, so much actually, now that I think about it, it's kind of a stupid question, but you couldn't even consider back when you were building things like the smart Wheeler and everything.

00:41:21:12 - 00:42:26:13
Unknown
I mean, it's evolved so, so much.

00:42:26:15 - 00:42:45:22
Unknown
Yeah. And it's all interconnected for sure. We're going to take a quick break, another quick break, and then when we come back a little bit more on agents and then I do, I, I got to ask you about sovereignty because we're kind of in a moment, especially after this weekend with anthropic models being removed and everything, it ties right into a topic.

00:42:45:24 - 00:42:51:19
Unknown
You talk about a lot around sovereignty. So we're going to take a break, come back and round things out with Joel.

00:42:51:23 - 00:43:15:11
Unknown
Okay. We'll chat a little bit more with Joel about the mythos five sovereignty question that I mentioned earlier. But first I just want to let you know we've got a YouTube channel, YouTube.com Inside Show. Or just go to YouTube and search for AI inside and the podcast and you'll find it. It's there. If you are listening to this interview and you want to see it in video, hey, that's where you get it.

00:43:15:12 - 00:43:21:19
Unknown
So go to YouTube, search for AI inside, show more with Joel coming up after the break.

00:43:21:21 - 00:43:44:14
Unknown
All right. We've definitely talked on the show a lot about about agents let's say agents behaving badly. We have the stories about like the coding agent wiping out the database happened not too long ago. The support bot story that happened recently that was essentially socially engineered into hijacking and rerouting accounts and stuff.

00:43:44:15 - 00:44:07:07
Unknown
You were on the 20 VC podcast last year. You said that the agent era version of hallucination is impersonation, which is super smart. You pointed at agents infiltrating like bank systems, that sort of stuff. And you called security a cat and mouse game, which I completely agree, and that we need to build rigorous standards around it. That was seven months ago.

00:44:07:11 - 00:44:30:20
Unknown
Cohere is shipping agents into financial institutions as we know what does what is North do today that stops an agent from taking an action that it shouldn't take, that it could, you know, maybe and maybe that's a protection that it couldn't do seven months ago, that maybe you didn't have the visibility on. But I'm curious to hear your take on that.

00:44:30:23 - 00:44:35:03
Unknown
Actually. Like.

00:44:35:06 - 00:45:52:05
Unknown
Okay.

00:45:52:07 - 00:46:24:05
Unknown
Yeah. So it's not just technology. It's it's also equally as important about the standards that are built around at that, like scaffolding around it. Yeah. Yeah. Interesting. Well, I did mention sovereignty and I want to make sure that we talk a little bit about that. Like I said now is there's a lot happening as of this weekend, three days ago, at least at the time of the recording of this show, this interview, the US government forced anthropic to cut off foreign access to fable five, to mythos five.

00:46:24:05 - 00:46:51:13
Unknown
Anthropic then basically chose to pull both models for everyone planet wide, let's say. And you know, it just kind of occurred to me, as I was preparing for this interview, that this is the scenario that your sovereignty kind of perspective is really built around. I'd love to hear your thoughts on kind of watching this happening from from the sidelines along with the rest of us.

00:46:51:15 - 00:48:46:17
Unknown
Yeah.

00:48:46:19 - 00:49:11:13
Unknown
Okay. So cohere for for those who are listening and don't know, Canadian, Canadian based or Canada based. So Toronto. Right. That's where you're where your HQ is. But there are some kind of us like tendrils. I don't know what the right word is. You've got you know, you've got US shareholders, some US executives, Capitol capital routed through Delaware.

00:49:11:13 - 00:49:39:09
Unknown
If is there any concern here with what's going on with with cloud and anthropic that, that you might not be insulated from such an event if it were to go that route?

00:49:39:12 - 00:49:42:14
Unknown
And.

00:49:42:17 - 00:50:39:11
Unknown
So they are.

00:50:39:14 - 00:51:04:11
Unknown
Yeah, 100%. I know we're we're running out of time. Wanted to kind of end with this. And this is a little bit of a of a kind of flashback. Not not quite 20 years, but while you were still at meta, you were on the Me Myself in AI podcast, and you had said then that you wished AI systems could understand each other, that so many of them are built for individual performance agenda.

00:51:04:11 - 00:52:16:00
Unknown
Collaboration was primitive and mind you, this was 2024 as we know. It's amazing how fast everything has developed in even a few short years. So a lot has changed in the field. Now, a few years later, are we any closer in your mind to agents that can build those connections? Like are you optimistic or kind of satisfied with where we are, or is there more room to grow their.

00:52:16:02 - 00:52:29:25
Unknown
Interesting?

00:52:29:27 - 00:53:22:12
Unknown
Where does the the that field need to go that we aren't putting enough attention to right now? Like what? Where does it need to go? Where do we need to put more of our efforts in our resources?

00:53:22:14 - 00:53:40:23
Unknown
Indeed. Joel, thank you so much for taking some time with me today. I'm so thrilled that I got the chance to chat with you for almost an hour. I really appreciate it, and I think you're doing great work at cohere and love to have you back sometime. Appreciate it.

00:53:40:25 - 00:53:42:22
Unknown
Thank you. Joel. We'll see you soon.