Skip to main content
MANIFOLD
Which new-to-me technical AI safety paper will I find coolest/most impressive?
2
Ṁ480Ṁ41
resolved Nov 9
50%Other

Time to see how well Manifold's crowd sourced search works.

If there's only one respondent but it is a technical AI safety paper I haven't heard of I'll still resolve to that.

I only have so many hours for reading papers so market may resolve days to weeks after the close date.

So people don't waste their time: I'm familiar with all the work put out by Anthropic, OpenAI, ARC, and MIRI, most-or-possibly-all of DeepMind's work, most of CHAI's work, and a tail of random other stuff. I generally find work more impressive as it gets closer to being about actual AI systems in the real world (submit papers about idk tabular RL at your peril)

Oct 4, 12:55pm: I'll also give tips for any papers I think are cool even if they don't win.

Market context
Get
Ṁ1,000
to start trading!
Sort by:

I'm not shocked nobody has posted any other papers after I threw so many up with high uncertainty, but I'm surprised nobody has even taken sides against me on any of these!

@L Well if no one else bets you're guaranteed to win I guess

@vluzko you haven't made me less interested in your library, though!

@L I'm an academic, I read a lot of papers. It would just take too long.

to others: I'm barely reading my suggestions - many of my suggestions are likely not great. some may be quite good. reducing my uncertainty may have significant profit.

if you were to post a list of papers you've read, it would be beneficial for others who want to read them, for avoiding posting ones you've already read, and for helping others walk the research graph from those papers.