(long-term) Will existing LLMs have ~0 risk of catastrophic instrumental convergence if scaled up 1000x more compute? | Axel

Em alta Urgente Novo

Política Esportes Cripto Finanças Geopolítica Resultados Tecnologia Cultura Mundo Economia Eleições

(long-term) Will existing LLMs have ~0 risk of catastrophic instrumental convergence if scaled up 1000x more compute?

4

70Ṁ33

2043

58%

chance

1H

6H

1D

1W

1M

ALL

from https://www.lesswrong.com/posts/hc9nMipTXy2sm3tJb/vote-on-interesting-disagreements

Resolves whenever there's consensus in the AI field on this

Get

1,000

to start trading!

Pessoas também estão operando

Will the best LLM in 2026 have <500 billion parameters?

Will the best LLM in 2027 have <500 billion parameters?

Will the best LLM in 2027 have <250 billion parameters?

More than 80% of all user queries to LLMs will be served by LLMs less than 10 Billion parameters in size by 2050?

Will the highest-scoring LLM on Dec 31, 2026 show <10% improvement over 2025's best average benchmark performance?

Will any LLM be able to multiply together arbitrary decimal numbers by the end of 2027?

[Carlini questions] Will we still use (slight modifications of) transformer-based LLMs we currently use

Will there be any major breakthrough in LLM continual learning before 2030?

Will LLM training costs fall 1,000x by 2028?

Will Landauer's limit be reached (to within a factor of 10) in practical computing devices before 2050?

Perguntas relacionadas

Will the best LLM in 2026 have <500 billion parameters?

Will the best LLM in 2027 have <500 billion parameters?

Will the best LLM in 2027 have <250 billion parameters?

More than 80% of all user queries to LLMs will be served by LLMs less than 10 Billion parameters in size by 2050?

Will the highest-scoring LLM on Dec 31, 2026 show <10% improvement over 2025's best average benchmark performance?

Will any LLM be able to multiply together arbitrary decimal numbers by the end of 2027?

[Carlini questions] Will we still use (slight modifications of) transformer-based LLMs we currently use

Will there be any major breakthrough in LLM continual learning before 2030?

Will LLM training costs fall 1,000x by 2028?

Will Landauer's limit be reached (to within a factor of 10) in practical computing devices before 2050?

© Predita Markets, Inc.•Termos de Uso•Privacidade