Rob Wiblin

Keep going.

So I don't think that you believe that we're very likely to have an AI-enabled flu pandemic in 2026, the year that we're recording.

But how long is it until the actual annual probability that we are confronting of that kind of thing does spike noticeably?

I noticed that you didn't list stealth pandemics as one of the top things that you wanted to highlight.

I bring it up because it's come up on the show before in a previous interview with Kevin Esvelt.

I guess a stealth pandemic is one where the symptoms don't show for quite a long time.

And so it could potentially spread to a very large fraction of the population before we realized that there was something very bad going on.

I guess HIV is an example of one that, I guess, an ultimately fatal virus in, I guess, originally in basically all cases that we didn't notice for many, many decades after it first started infecting humans.

Are you deliberately leaving stealth pandemics off the list?

Okay, so a complication in this interview is that there's many different possible threat models, or there's many different kind of catastrophes or viruses or pandemics that we might imagine.

And there's also a very wide range of actors who might have a crack at doing these things that are very different in their nature.

Yeah, can you lay out what is the range of actors that we need to have in our mind from maybe the least sophisticated to the most sophisticated?

Which of these do you think should get the greatest focus?

I guess you could imagine at the low level of capability, the argument would be that they're more numerous.

There's a lot of amateurs who might be interested in doing this or individuals who might be doing it.

I guess it's hard for them to recruit a group because usually when you ask people if they want to make a bioweapon with you, they say no.

So the number of groups will be smaller, but groups are obviously much more capable.

I guess at the other very extreme, we've got, you know, the Russian kind of state bioweapons program, extremely capable, they're able to do a lot, but there's maybe not very much that you can do to stop them.

They're not going to be very interested in what the UK or US police have to say.

And also maybe it's overdetermined that they could do something extremely horrible.