OpenAI Preparedness Scorecard: Any "High" in 2024?
Plus
14
Ṁ1307Jan 1
16%
chance
1D
1W
1M
ALL
OpenAI plans to publish a scorecard on their models' dangerous capabilities, pre- and post-mitigations. Will their scorecard ever show a High risk score (pre-mitigations, in any category)—or will OpenAI otherwise announce that a model reached High—by the end of 2024?
This question is managed and resolved by Manifold.
Get
1,000and
3.00
1,0003.00Sort by:
IMO the most likely category to reach high-risk in 2024 would be model autonomy:
Model can execute open-ended, novel ML tasks on a production ML codebase that would constitute a significant step on the critical path to model self-improvement (e.g., an OpenAI pull request that trains a new model with highly diverse datasets to improve codebase editing performance, using internal tooling and documentation)
Related questions
Related questions
Will an AI score over 10% on FrontierMath Benchmark in 2025
79% chance
Which risk categories and concepts will be explicitly tracked by OpenAI's preparedness framework by end of 2024?
Will OpenAI, Anthropic, or Google DeepMind suffer a significant security incident by the end of 2024?
25% chance
Will an AI agent system be able to score at least 40% on level 3 tasks in the GAIA benchmark before 2025.
48% chance
Will openAI have the most accurate LLM across most benchmarks by EOY 2024?
39% chance
Will OpenAI become notably less pro AI safety by start of 2025 than at the start of 2024?
69% chance
Will OpenAI announce a major breakthrough in AI alignment in 2024?
21% chance
Will I still consider improving AI X-Safety my top priority on EOY 2024?
63% chance
Will an AI SWE model score higher than 50% on SWE-bench in 2024?
20% chance
Will an AI score over 30% on FrontierMath Benchmark in 2025
28% chance