The Exchanges

Q Yeah. Ben, I have to ask you, what is your hottest take right now?

A My hottest take right now is that I think AI is gonna converge, and, ah, we are going to build a digital twin of nature, and just like we have prediction models for, like, everything from, like, missile defense to, ah, tsunamis and earthquakes, we're gonna have that for nature, and we're gonna understand in real time, like, how nature is affected by every decision that we make, and that's gonna be Incredibly enlightening and terrifyingly scary, but I think we're going to build a digital twin of nature, we as humanity, uh, where we will be able to truly understand the consequences of our decisions, and then we will be able to leverage technologies like what we have to productionize species to combat some of the adverse effects of those decisions.

AI assessment note: “My hottest take right now is that I think AI is gonna converge”

Q that require real expertise and real-world judgment. That's why companies like NVIDIA, Anthropic, Salesforce, and Gemini partner with Turing. Turing builds realistic reinforcement learning environments and data systems based on real operational traces, the kind of infrastructure Frontier Labs need to train superintelligence. Visit Turing.com slash S-O-U-R-C-E-R-Y. So to set the stage, you started this on the premise of extinction. So what are the stats? I know this is dark.

A No, it's super dark. It's like, it's like, but, but like, it's dark, but I think it's hopeful, right? Like I'm, I'm an eternal optimist. It's currently forecasted that we're going to lose, uh, up to 50% of all biodiversity in the next 25 years. So that, that's dark. I mean, it's, it's, it's, yeah, it's, but it's like, that's why we have to do stuff, right? And so, um, And so for us, if we can get everyone, like, sometimes people get confused that, like, we think that de-extinction is a replacement for modern conservation. We do not think that. But the reality is, is like, no matter how good conservation works, it just doesn't move at the speed of which we're eradicating species and changing the planet. Like, we're doing that. Like, that is happening. So for us, uh, if you have a combination of a de-extinction toolkit, That has lots of other ramifications, which we've talked about some, uh, and you have a glo, you get governments to invest in a global distributed bio vault network. So they're at least backing up things. At least we're starting to build a backup plan. And then lastly, we open source all of our technologies for conservation. So anybody can use any of our technologies for free for conservation.

AI assessment note: “we're going to lose, uh, up to 50% of all biodiversity in the next 25 years”

Q So you guys do, between that and the labs and how everything is designed here in this secret location, how did you get, I mean, you have a very distinctive style here. Like, how did you guys develop that style and the branding on that?

A So we wanted to make, once again, Couldn't be more underqualified. We wanted to make, um, we wanted to make, this kind of goes back to inspiration. We wanted to make science cool and fun and engaging. And so when we set out to do the, the task of like getting rid of extinction, uh, we wanted, you know, we thought it was a big problem. So we thought Colossum was a good name because it's a pretty big problem that we're trying to solve. Um, and then at the same time, we're also working with cells, which we thought was funny that we're working the most, Small things, but hopefully they're successful to make really big things. So we're kind of planning success, right? With that, with mammoths and stuff. So, so that's where we came up with the name. But then we had this idea that it's like, if you're going to go back to the like Bill Nye old school models of like science and excitement where it's like, you're putting in the red food coloring and then shit's exploding in the volcano versus like today where it's like only spelling bee champions and like kids inventing like nuclear fusion. If you want to go back to, like, the stuff that I could do, like, in, in, like, the eighties and nineties, well then, like, that's, like, old school MTV, like, back before it was, like, 16 and pregnant, and it was, like, TRL, like, how do you go back to, like, the TRL days of MTV and mash that up with…

AI assessment note: “how do you take, like, Harvard meets MTV? That was, like, Our thesis”

Q Yeah. So how did you guys prove them all wrong?

A I think that, you know, trust, I think it's easy to raise capital when you have trust. And I think trust is built with setting and managing expectations. So we've, I think done a really good job with our investor base and with hopefully the world, uh, and also, uh, our board of setting and delivering on expectations. And I think we, we've shown, we've had these surprise, like the Willie Mice for surprise and delight moment, The direwolves were definitely a surprise and delight moment. Um, and so, so I think that, uh, like delivering on, on kind of our, uh, roadmap, we've been very consistent on kind of our delivery and we've been ahead. And then I think we've also done a really good job of exceeding those expectations of what's possible. Like we were told if in five years we were doing five to seven edits, we'd be great. We're now doing hundreds. You saw I left. We're doing hundreds of edits. We, we haven't done this yet, but we're now pushing Uh, a new construct that we've designed and a new delivery mechanism that we think we can be doing thousands of edits at once. Like these are things that like, you know, a year, like three years ago was like sounded impossible. A year ago still sounded impossible. Right. And so we, I think we're, we're consistently showing that. And so when you build that trust, I think it's very easy for people to, to continue to deliver or to, to invest…

AI assessment note: “we were told if in five years we were doing five to seven edits... now doing hundreds”

Q So do you get like a report at the end of the day, at the end of the month? Like how do you learn and train off

A Yeah. So on that data, that's highly visual data. So, uh, so it's all, um, screen capture and can, uh, computer, uh, vision, uh, and from an input and capture perspective, we track efficiencies and cloning rates. And, and, and so we, we track all of that in spreadsheets. Uh, now with thanks to like some of the next gen applications of, of AI in the early days, things like open AI and Claude and others, uh, Um, more super helpful to our company because it's like, turns out that like, you know, Opus doesn't do like an ancestral state reconstruction of like a genome and like it doesn't like build comparative genomics models. So we had to go build all of that, which I think was good because we had to go build up an AI team and think through some pretty hard problems. But now we're connecting existing systems like You know, most labs don't use things like JIRA, and so I think the hardest thing we've done is reprogram, not stem cells, but scientists to work in things like JIRA and Smartsheets, like traditional software, like things, and then, and then we're connecting those to like the lab notebooks and some of our proprietary systems all through an AI layer, right? And so the visibility that we can give our mid-level and senior management on the science teams of what's happening in the lab In near real time through, ah, the aggregation of AI across systems. Now we're implementing so…

AI assessment note: “visibility that we can give our mid-level and senior management... in near real time”

Q that I saw when we were walking through the lobby was with the dire wolf. You see this, like, this, you guys have this, like, little, like, model of, like, how land has evolved over time, and the benefits of the dire wolf on the land, and how biology changes, and all this kind of stuff. So, like, in practice, Are you going to be releasing any of these animals?

A Yeah. So in a couple of phases. So obviously you have to work very closely with governments, indigenous people groups, private landowners for our extinct species. We do want to put them back into the wild, uh, with all these collaborations in place, right? Where they help the ecosystem, um, the government supportive, indigenous people groups are supportive. We want to do that for our, for our, Gene drives and whatnot, that's also highly up to the government, right? Like right now we are educating, uh, USDA, the intelligence community, the Department of War and others on these technologies and how they can combat these issues domestically and abroad. So, uh, but it is up to them, right? Like they have to give us the green light. Um, like there were some, there were some, I did an interview about the screw worm, then it went viral. And there were, like, 20 articles about it, and then there was feedback on, like, Blue Sky of all places. Surprise, there was negative feedback on Blue Sky, right? The echo chamber, you know, or liberal Twitter, right? And so, uh, there was feedback on Blue Sky that's like, well, Colossal doesn't have permits. Like, yes, we do not have permits, but, but this is a technology that, if it is permitted, can eradicate this problem. And so, like, in the screw worm example, uh, it's like, they're spending a 1,000,000,005 in taxes, and in eight years, they're …

AI assessment note: “We do want to put them back into the wild”