Will Scott Aaronson be at least as optimistic about AI alignment next year?

Ṁ360Ṁ1.3k

resolved Jul 3

Resolved

YES

ALL

Scott Aaronson just announced that he's working at OpenAI for the next year: https://scottaaronson.blog/?p=6484. In this post he also says "just as Eliezer became more and more pessimistic about the prospects for getting anywhere on AI alignment, I’ve become more and more optimistic". Resolves YES if based on Scott Aaronson's blog posts or other statements after about 1 year, he generally is at least as optimistic about AI alignment as he is today; NO if he's less optimistic.

Market context

Technology

Get

1,000

to start trading!

🏅 Top traders

#	Trader	Total profit
1		Ṁ128
2		Ṁ62
3		Ṁ45
4		Ṁ31
5		Ṁ26

People are also trading

Will some piece of AI capabilities research done in 2023 or after be net-positive for AI alignment research?

81% chance

Which Scott Aaronson AI world will come to pass?

AI honesty #2: by 2027 will we have a reasonable outer alignment procedure for training honest AI?

24% chance

Conditional on their being no AI takeoff before 2050, will the majority of AI researchers believe that AI alignment is solved?

52% chance

Conditional on their being no AI takeoff before 2030, will the majority of AI researchers believe that AI alignment is solved?

34% chance

Will ARC's Heuristic Arguments research substantially advance AI alignment before 2027?

26% chance

Which Scott Aaronson AI world will come to pass? (Metaculus forecast, 2027-12-31)

By 2028, will I believe that contemporary AIs are aligned (posing no existential risk)?

33% chance

Will a Turing Award be given out for work on AI alignment or existential safety by 2040?

79% chance

Is AI alignment computable?

Sort by:

It's getting close to the 1 year mark. What do people assess about how optimistic Scott is on AI alignment recently? He's talked about the topic a lot (example post https://scottaaronson.blog/?p=7230 ) but I haven't been following his posts closely so I'm not sure, if anyone else has a sense let me know.

AI safety: what should actually be done now?

So, I recorded a 2.5-hour-long podcast with Daniel Filan about “reform AI alignment,” and the work I’ve been doing this year at OpenAI. The end result is … well, probably closer …

I still don't know how to resolve this - he's posted some long interviews about AI safety recently but I don't have time to listen to them.

If traders could provide evidence/info/data, that would be helpful. Especially the people who traded recently - @KatjaGrace you seem quite confident.

predictedYES

@jack I just asked Scott and he said "I’d say I’m about the same amount of optimistic as a year ago — so that’s a yes I guess".

@KatjaGrace Thank you!

Look forward to someone who can’t possibly grasp the basics of Roman, German, or current political events being assigned to defeat a superhuman AI

(Once again, send in someone with just the least bit of practical real world knowledge not the egg-heads)

There are exactly two things with vastly asymmetric attacker-defender dynamics: mind viruses and biological viruses.

Based on his physiognomy and npc-level-0 he’s unfit to defend against AI deploying either

Tfw someone’s blog seems so smart until Gell-Mann amnesia hits and you realize he has the historical knowledge of a six year old and the temperament of a crazy cat lady addicted msnbc and boxed wine