What will be the new "move 37" for AI/LLMs trained with RL?
1
100Ṁ42030
60%
Discovery of novel concept in mathematics
50%
Design of novel and widely useful electronic component
50%
Design of novel and widely useful mechanical component
50%
Proposal of new economics model
50%
AI proposes a convincing explanation about about a previous event
Resolves on January 1st 2030 to the set of answers which are analogous to "move 37" in the context of AI and/or LLMs trained using reinforcement learning. These answers must describe emergent behavior in AI/LLMs made public knowledge after January 1st 2025, which become topics of significant interest and debate (similar to the discussion over "move 37").
Inspired by this tweet from Karpathy:
This question is managed and resolved by Manifold.
Get 
1,000 to start trading!
1,000People are also trading
Related questions
Will the most interesting AI in 2027 be a LLM?
70% chance
Will the next major LLM by OpenAI use a new tokenizer?
77% chance
Will Transformer-Based LLMs Make Up ≥75% of Parameters in the Top General AI by 2030?
50% chance
Will RL work for LLMs "spill over" to the rest of RL by 2026?
34% chance
Which High-risk threshold as defined by OpenAI will be reached first by an LLM, whether or not that LLM is released?
What will be true of OpenAI's best LLM by EOY 2025?
OpenAI to release model weights by EOY?
84% chance
There will be one LLM/AI that is at least 10x better than all others in 2027
17% chance
Will an AI Tutor (LLM personalized for each student) replace conventional teaching by 2035?
67% chance
In Jan 2027, will the top 3 leading AI labs be offering fewer than 15 flagship LLMs between them?
61% chance