Will Eliezer Yudkowsky work for any major AI-related entity by 2027?
Standard
26
Ṁ757
2026
20%
chance

Eliezer Yudkowsky has been a leading voice in AI alignment for years, but has never done any work directly for a major AI company or organization.

Will that change by 2027?

Get
Ṁ1,000
and
S1.00
Sort by:

(MIRI is a big name tho)

How about if Paul Christiano manages to boot up a huge effort within the government with regulatory power over the labs, and brings Eliezer in on it?

Paul is currently the head of the US AI Safety Institute as part of NIST - I wouldn't currently consider them "major" since they don't seem to have a lot of regulatory power - but the US government does have weight to throw around and that may be a channel through which that weight flows at some point.

And, additionally, Paul Christiano might have a "oh crap, Eliezer was right" moment at some point.

Idk if thats cope - but I think Paul is sufficiently strong to eventually locate what world he's living in, and I think we are living in an Eliezer flavored world.

To me, the value of the probability estimate this question provides is "Is Eliezer effectively in charge of an effort that has a shot at acquiring ~2+ years of lead time - or otherwise dictating the direction of whichever set of entities in the field have and can hold that amount of lead time over the rest of the planet."

That... Probably won't actually happen and probably would not be enough to save us if it did, but if it did happen maybe we have a shot in the worlds where there's a miracle that gives us some kinda plausible shot at flipping the game board?

I may just be too tired right now, but I have this mental image of humanity trying to thread a strand of spiders silk through the eye of a needle - but we don't actually know if we'll have the needle in hand in the near future, and the thread of spiders silk is being blown around in the wind in a field 2km away.

Like, Alan Turing needed to have a slightly better life, and write a huge book describing the abstract argument for the alignment problem, and John Von Neumann needed to fill in all most the details re. basic mind design, and everyone in the AI field needed to be taught that stuff on day 0 of their careers - and maybe we needed to also not put lead in gasoline or hollow out a bunch of populations with all the death and misery in the 20th century...

Okay, I'm getting off topic.

I sorta suspect that 19% is too high. People in charge of these companies don't seem to appreciate Eliezer or think it would be a good idea to elevate his status.

If "the US government/Paul Christiano makes him defacto US AI development micro-manager" counts as YES - then maybe there's a 5% chance?

What qualifies as "major"?

Would being advisor or something like that count

@Asher no, unless his role as advisor is very extensive and significant.

I'm supposing his current job doesn't count?

@NivlacM that is?

@firstuserhere founder of and a research fellow at the Machine Intelligence Research Institute (MIRI)

@NivlacM I dont think MIRI is a major AI related entity

Similar for shorter timeline: