00:23:48 · Tom McGrath
[00:23:48] S1: the plug. I try. So, yeah, it was definitely, like, a bit of a wiggly path. And what we settled on is this idea of Silico, and Silico is a a way of, like, empowering people to do interpretability and model AI research. You know, if you open up Silico, what you get is essentially, you can talk to an agent and it will kind of go off and use... Either use our libraries of interpretability and training, software that we spent a lot of time on and, you know, the sort of deep skills encoded in there from all the kind of collective wisdom of our researchers. And you can use that to do whatever AI or, you know, research. Like, we're specializing in interpretability, but you can also do training in there. You can also do, like, data research in there, that sort of thing. And the idea here is, like, if we went out and we individually taught everyone to do interpretability, we would just never, like, get enough of it done. Like, as you say, as as increasingly capable of models are coming out and people want to own their own own their own models, they are going to want more complicate... Like, more powerful training services. They won't gonna want want to be able to, like, understand the model they're training, control it during training, like, you know, for instance, like this predictive data debugging stuff, they... They've they've gonna have a load of their own data. Some of it is gonna be great. Some of it is gonna, by surprise, teach the model all sorts of things you didn't want it to teach. And so you're gonna want ways of, like, curating that data, debugging individual, like, rollouts from production back into the data or into the model and saying, like, how can we fix this? That sort of thing.
00:25:27 · Aditya Agarwal, Tom McGrath
[00:25:27] S0: Yeah. I mean, it kinda gets to... It's a useful segue to the the last thing I wanted to talk about, which is kind of research generally. Right? Like, you're almost talking about... Or the way you put it to me, I think, before we... When we were chatting earlier is, you know, it's it's not exactly correct to call it, like, a research platform. But also, you know, like research and product have become so kind of fuzzy, right, like with AI. And you even see this in the way that we sort of talk about certain types of companies. Right? Goodfire is a research lab. Right? The lab alone. Right? Like, the the, foundation model labs or that. Right? Everyone uses that term. But you... You've had kind of an interesting perspective as someone who has experienced research. First in academia, when you got your PhD, you were at ICL. Right? [00:26:14] S1: Yes.
00:26:14 · Tom McGrath, Aditya Agarwal
[00:26:14] S1: Yes. [00:26:15] S0: And then then you were at DeepMind. Mhmm. So you're, you know, kind of the the big tech version of, of an AI lab. And then you cofounded a company. Right? And and you're the chief scientist at a at a, small but growing start up. How has your relationship to the idea of doing AI research changed as you've moved through those? Because I I meet a lot of people who are coming from research backgrounds and are thinking about making similar sorts of transitions. Right? And in fact, very kind of weirdly, I've sort of ended up specializing in working with researcher founders because my background is in, like, storytelling. Mhmm. And it turns out that's, like, a very value... That's a particularly valuable skill when working with researcher founders. [00:26:58] S1: Yes. We we can be very, like, literal. [00:27:00] S0: Yes. Yeah. Right. Exactly. Kinda learns a little bit of the the, like, figurative magic. But also it's just, like, often kind of the most ambitious and sci fi stuff. Right? So, like, I I have an affinity to it. But I, you know, I often find myself talking to people about that that process of making the jump from kind of one version of those to the other and and and... Like, because of where I am at SPC to start ups. But, like, how... What has your experience been like making those transitions, and and how has your... How has research changed for you?
00:27:28 · Tom McGrath
[00:27:28] S1: Yeah. The one thing that hasn't changed is I still love it. Like, I I just think that doing doing science is just, like, one of the... Kind of almost like one of the core human activities. You know? It's sort of... There are not that many things that feel more kind of human than doing... Than trying to understand what's going on. You know, you could imagine that that spent, like, tying your life so closely to something like that would kind of kill your passion for it. But the thing that hasn't changed is, I guess, I just love it. And some of that is to adjust interpretability being so much fun. I suppose I've been going gradually towards more kind of needing to be pragmatic while still trying to maintain this, like, long term vision. And it's remarkable, the effect of, like, having a deadline. Having a deadline is just, like, actually quite quite good. And and sometimes you need to organize things and that kind of thing. I think this may be a personality thing. Obviously, I'm talking about distribution here, but there'll be outliers in either direction. But, like, I think most... More scientists tend towards being kind of relatively egalitarian. They're like, well, we'll just kind of have... We'll kind of organize it all out to ad hoc, which I think is a nice tendency.
00:28:38 · Aditya Agarwal, Tom McGrath
[00:28:38] S0: I do like the bit about about the deadline. I mean I mean, there is sort of like a practical relationship that just changes to the work. Right? Like, the point of research in in a company is is... Hopefully, it eventually translates through to economically valuable product. Right? The degree of distance from that ultimate product [00:28:56] S1: Yeah. [00:28:56] S0: Sort of is inversely, related to, like, the size of the organization, or is that not the case? [00:29:05] S1: Interestingly, I think it's actually got smaller as we've grown, which is kind of funny. I think perhaps the reason... Maybe there's a sort of confounder here where as we have... We now have this product surface, Silico, where it's, like, very easy to translate stuff. And so that means that rather than there being some sort of relatively long long process of, like, we we have some... We maybe get an idea as a result of some some consulting agreement or some some, like, problem that a customer has with some particular product that we try and develop. We just, like, put the research in there, and it either works or it doesn't. And so that that sort of... That gets rid of a lot of the organizational complexity. So even though we've grown, I think, actually, like, the contact between research and kind of commercial reality has gotten much smaller. I'm not sure if there's something generalizable there, but it seems pretty nice.
00:30:00 · Aditya Agarwal, Tom McGrath
[00:30:00] S0: I mean, it it sort of follows, you know, like, in the sense that if you are a, research oriented startup, if you were a research lab, however you, you know, characterize it, early on, sort of definitionally, the bet is you're trying to figure out some core things Yeah. That should be the foundation for, you know, like a a a product or a series of products or things like that. Right? Which is to say that you have to do that work first. Right? Like, investing into that more... That that that is the core kind of, like, risk Yeah. Early on. Right? It's not, oh, you know, can we build, the... Or, like, do customers wanna pay for this product? So you you don't even know if it works yet. Right? It doesn't... It... Right? It is, the bulk of the risk is in, the sort of open technical or, research questions. [00:30:46] S1: Yeah. [00:30:46] S0: So you would expect that... Yeah. Or, I mean, this is what you see at OpenAI or Anthropic. Right? For years, they were a bunch of researchers and then figured some stuff out, implemented research from the rest of the community, and, like, started to to generate product. Right?