
Will AGI figure out a quantilizer of which people roughly continued most to its global computation?
Mini
2
แน302060
30%
chance
1D
1W
1M
ALL
Get แน1,000 play money
Sort by:
@stochasticcockatoo Can you please extend this market? I see you have been active. So asking before I extend it or another mod.
@SirCryptomind I think the market should be extended instead since there isn't (as far as I can tell) an unambiguous "AGI" yet.
@cockathiel is this specifically by the end of the year, or just in general?
(ChatGPT says it doesn't know what kind of people contributed to its training data.)
Related questions
Related questions
Are LLMs capable of reaching AGI?
59% chance
When will there be an AGI for most tasks I care about?
Will we reach "weak AGI" by the end of 2025?
33% chance
Will we have an AGI as smart as a "generally educated human" by the end of 2025?
57% chance
Will AI come up with a 99.99 percentile quintilizer for most perceptive people in the planet by 2030
41% chance
Is slowing down AGI good for AI safety? [resolves to poll]
83% chance
Will we have strong AGI by 2030? (metaculus criteria)
27% chance
Will AGI be maximally impatient in trying to expand its causally-affectable portion of the universe?
42% chance
Will AGI figure out who is the ideal integrator of the conscious human experience by 2030
23% chance
Will Q* be legally considered AGI?
10% chance