TurnTrout et al will claim they found a love vector which qualitatively modifies completions in a range of situations

#AI Alignment #Algebraic value edits

5

5

Ṁ48

Ṁ130

resolved May 13

Resolved

YES

1D

1W

1M

ALL

https://www.lesswrong.com/posts/gRp6FAWcQiCWkouN5/maze-solving-agents-add-a-top-right-vector-make-the-agent-go

Resolves based on follow-up post

Apr 2, 11:37am: ~~They will claim they found a love vector which qualitatively modifies completions in a range of situations~~ → TurnTrout et al will claim they found a love vector which qualitatively modifies completions in a range of situations

Get Ṁ200 play money

🏅 Top traders

#	Name	Total profit
1		Ṁ15
2		Ṁ1
3		Ṁ0

Related in AI Alignment

What will be announced by OpenAI during Monday's livestream? [ADD RESPONSES]

By the end of 2026, will we have transparency into any useful internal pattern within a Large Language Model whose semantics would have been unfamiliar to AI and cognitive science in 2006?

See more AI Alignment questions

Related in Algebraic value edits

Algebraic value editing works better for larger language models, all else equal

Will someone find a truth-telling vector which modifies completions in a range of situations by 2024-10-24?

See more Algebraic value edits questions

More related questions

Eliezer Yudkowsky is impressed by a machine learning model, and believes that the model may be very helpful for alignment research, by the end of 2026

ChatGPT (Or LLMs really) have discovered regularities in language that humans are not aware of

Between 2030 and 2035 there will be evidence of different models of neural network collaborating better with eachother

Has manifold mostly: (A) nerdsniped those who could be doing important research, or (B) helped people make predictions

A new image generation system comes out which has accurate text generation, and interest in it surpasses Stable diffusion and Midjourney

In 2029, will any AI be able to construct "reasonably" bug-free code of >= 10k LOC from a natural language specification? (Gary Marcus benchmark #4)

Suspiciously specific predictions

AI Alignment questions

What will be announced by OpenAI during Monday's livestream? [ADD RESPONSES]

By the end of 2026, will we have transparency into any useful internal pattern within a Large Language Model whose semantics would have been unfamiliar to AI and cognitive science in 2006?

Algebraic value edits questions

Algebraic value editing works better for larger language models, all else equal

Will someone find a truth-telling vector which modifies completions in a range of situations by 2024-10-24?

Related questions

Eliezer Yudkowsky is impressed by a machine learning model, and believes that the model may be very helpful for alignment research, by the end of 2026

A new image generation system comes out which has accurate text generation, and interest in it surpasses Stable diffusion and Midjourney

ChatGPT (Or LLMs really) have discovered regularities in language that humans are not aware of

In 2029, will any AI be able to construct "reasonably" bug-free code of >= 10k LOC from a natural language specification? (Gary Marcus benchmark #4)

Between 2030 and 2035 there will be evidence of different models of neural network collaborating better with eachother

Suspiciously specific predictions

Has manifold mostly: (A) nerdsniped those who could be doing important research, or (B) helped people make predictions