Now what, what can you tell me about this particular experiment, and more importantly, what are the implications of this experiment for the, you know, biosecurity world?

Yeah.

So first, huge kudos to the scientists who did it because it was a big accomplishment.

I wasn't expecting it for another year and a half.

And number two, because they recognized it would be a problem.

So they did not train their AI model on not just human viruses.

They banned all viruses from animals.

Um, and, and it doesn't seem that we're so far away from someone being like, "You know, let's just go wipe us all out again." Walk me through exactly what you told Congress, and h- then in the year or two years that they've had, how, how have they totally fixed this problem?

Are we just lucky or are these things really difficult to engineer?

So first you ask really good questions.

My answer is of publicly known agents today, Setting aside like traditional bioweapons, the anthrax and so forth, which requires a lot of technical knowledge to produce, to purify and to deliver.

What you're, I think, really talking about is why haven't we seen a deliberate pandemic today? And I think the answer to that one is when it comes to credible will cause a pandemic agents available out there that we know that is just broadly understood, this will cause a pandemic.

There's really just one.

and that is smallpox.

We're not confident that any other agent would definitely cause a pandemic.

Nope, not a skin cancer

Not quite

... 'cause they trained it on 15,000 viruses that normally infect this kind of bacteria.

So it's more like imagining if you had trained it on all of the COVID variants, then the AI would spit out new COVID variants, and some of those would spread.

Now, that's a harder problem, but that would be the next phase.

And thankfully, right now, due to new government policy, you cannot get funding to run that experiment to actually make the predicted human viruses in the lab.

That's no longer allowed, and I think that's a good thing.

Mm-hmm.

Didn't train their broader AI on any sequences from human-infecting viruses.

So what's your view? What does raised by AI mean for you?

I think a key job for us is looking ahead at where technology as a whole is taking us, and I don't just mean AI.

AI is an accelerant.

It's the meta technology that can accelerate all of the others, especially cross-domain advances.

And it may well be that AI can be the equivalent of a... child safety seat and eventually seat belts.

When we grow up and we pilot fast-moving vehicles, we still wear seat belts.

And I don't know about you, but even when I get in a Waymo, I still wear a seat belt.

[laughs]

Mm.

Which means it's hard for these communities to say, "Actually, thanks, but no thanks." And most people in malaria-afflicted communities, I mean, you can imagine more than half a million children dead every single year, and you can just keep going back and back and back, and that just kept happening.

And people are people everywhere.

It's a tragedy when kids die, and they would love to have it stop.

The question is, do they trust this as a thing that can stop it, and do they trust the people proposing to do this to actually keep their word and tell them the truth?

So why not tamper carefully but broadly?

I envision a world that is kinder and more elegant, one where instead of spraying our fields with insecticides to kill the insects so they don't eat our crops, we instead engineer them to not like the taste, and otherwise go about their normal ecological role.

Where we can create new diversity, new forms of life, to solve problems that we're experiencing or that other organisms are experiencing.