Capabilities · Concept
Automated Multi-Model Review Benchmark for AI Scientist Systems
Protocol using frontier LLM reviewers to score autonomous research-generation systems on originality, rigor, clarity, and significance.
Forecast horizon
Connections
Connections · 2
How this node ties into the rest of the map, and the evidence behind each link.
to · references 4/10
Rated Conceptual Arguments Dataset for Hard AI/Philosophy Questions Both advance evaluation of AI systems on open-ended scientific/conceptual quality rather than solely verifiable answers.
+4 growth from · requires 3/10
Three-Stage LLM Pipeline for Major Mathematical Conjecture Discovery Autonomous conjecture and research pipelines need reliable evaluation of originality and rigor beyond formal parse checks.
+3 growthSignal sources
Signal sources
Dated facts from primary sources in this direction.
Task horizon doubling Mar 2025
The length of software tasks AI agents can do autonomously at 50% reliability has doubled about every 7 months — and since 2024 closer to every ~3 months.
METR → Benchmarks saturating Apr 2025
In one year scores rose by 18.8, 48.9 and 67.3 points on MMMU, GPQA and SWE-bench; real-world software solve rate jumped from 4.4% to 71.7%.
Stanford HAI — AI Index 2025 → Autonomous coding 2025–2026
On SWE-bench Verified (500 real GitHub issues), autonomous coding agents reached ~80–86% by late 2025, up from under 50% in early 2025.
Epoch AI →