Chris Strahle

Biosecurity specialist @ BlueDot Impact

[Chris Strahle @chrisstrahle_]

I expect that many of the best candidates for biosecurity talent work, much like myself, really enjoy technical work.

But being the technical person that orients exceptional people in the field may well be the most impactful thing for you to do.

https://t.co/4Cvx6NEYk3

[Chris Strahle @chrisstrahle_]

Great read!

I would add that although today's general-purpose AI and biological AI models likely don't provide a massive uplift, the market clearly expects this to change in the future, with both biopharma companies and frontier labs making big bets on AI in drug discovery.

https://t.co/8uvv8j2Cm8

[Chris Strahle @chrisstrahle_]

Wild to see the Lancet Commission listing malicious use of AI as one of the two threats that have the potential to lead to human extinction by 2100 (#2 is nuclear conflict)

https://t.co/WMDKGSDjXW

[Olivia H. Scharfman @OliviaHelenS]

I've been thinking hard about modern biosecurity for five years.

In the course of that thinking, I've learned a lot about what the field consensus is on a complete biodefense stack for the AI age.

Today, as part of my work with @BlueDotImpact, @chrisstrahle_ and I have published an overview of this defense stack, alongside a starting list of ten projects we would like to support motivated founders to build.

Funding will be available through the Seed Grants program.

Please share this with the best founders you know.

https://t.co/UpSxY4KwPo

[Olivia H. Scharfman @OliviaHelenS]

I had a very public argument with @DavidRBellamy.

Afterward, we did something rare online: we kept talking privately.

We discovered there is a lot we agree on:
- Biological risk is real, old, and tractable
- It is not mainly an AI-policy problem
- It is our job to make bioweapons a bad bet for adversaries
- There are clear, practical steps to increase biosecurity

I suspect this is true for many areas of public disagreement –– people may diverge on their philosophical opinions a lot more than on practical recommendations.

Sometimes, taking things private can lead to much better outcomes. Incentives in the public square are backwards. They reward winning over finding agreement.

As a society, we should have more adversarial collaboration to figure out what practical steps we agree on.

Here's ours:
https://t.co/WX1cNmb4jl

[Dewi Erwan @dewierwan_]

One of my biggest fears in AI safety is that this might be true:

> The vulnerable world hypothesis is likely correct. Once enormous amounts of cognitive labor start getting applied to basic science, it will quickly become apparent how many avenues exist to create cheap, ultradestructive weapons technology. Solving this problem without global preventative policing (e.g. AI nonproliferation) is impossible, because hardening civilians against all avenues of attack is too expensive and will take too long.

[Joshua Teperowski Monrad @jtmonrad]

"Conviction" is rapidly approaching "cracked" among overused hypewords, but there really is something to it

My latest (year-old, but now public) essay:

[Naval @naval]

If AI is risky like fire is risky, then you want everyone to have it.

If AI is risky like a nuclear weapon is risky, then you want no one to have it.

[Chris Strahle @chrisstrahle_]

Human biology is a lot harder to patch than software and AI will not speed that up significantly.

The design stage of the COVID-19 vaccine (which AI would speed up most) took ~2 days, confirming that it works took ~340 days.

Great post by @Noahpinion