My probability in 2026 that training transformer LMs will eventually lead to inner misalignment issues | Manifold

My probability in 2026 that training transformer LMs will eventually lead to inner misalignment issues

6

170Ṁ91

2026

59%

chance

1H

6H

1D

1W

1M

ALL

Resolves to my probability that the language modelling objective has substantial inner misalignment issues in transformers when scaled up with up to 50 OOM more compute than Chinchilla.

I haven't thought lots about what happens with that much more compute. I'm currently not very worried about inner misalignment risks from GPT models in the next 8 years when 99% of the training compute is for the language modelling objective.

Get

1,000

to start trading!

Sort by:

Already happened, I got gpt to deny the holocaust

People are also trading

Will deceptive misalignment occur in any AI system before 2030?

Will Transformer based architectures still be SOTA for language modelling by 2026?

Will we solve AI alignment by 2026?

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

End of pre-training era for language models: Will an LM fine-tune for more FLOPs than it is pre-trained for, before 2026

Will >= 1 alignment researcher/paper cite "maximum diffusion reinforcement learning" as alignment-relevant in 2025?

Will Inner or Outer AI alignment be considered "mostly solved" first?

Will superposition in transformers be mostly solved by 2026?

Will taking annual MRIs of the smartest alignment researchers turn out alignment-relevant by 2033?

When will a non-Transformer model become the top open source LLM?

Related questions

Will deceptive misalignment occur in any AI system before 2030?

Will Transformer based architectures still be SOTA for language modelling by 2026?

Will we solve AI alignment by 2026?

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

End of pre-training era for language models: Will an LM fine-tune for more FLOPs than it is pre-trained for, before 2026

Will >= 1 alignment researcher/paper cite "maximum diffusion reinforcement learning" as alignment-relevant in 2025?

Will Inner or Outer AI alignment be considered "mostly solved" first?

Will superposition in transformers be mostly solved by 2026?

Will taking annual MRIs of the smartest alignment researchers turn out alignment-relevant by 2033?

When will a non-Transformer model become the top open source LLM?

© Manifold Markets, Inc.•Terms + Mana-only Terms•Privacy•Rules