Development

InnovationEval finds limited recovery of a human research breakthrough

The author treats the result as a counterpoint to claims that AI research is already automated.

development2026-10-12benchmarks

Claim

The newsletter reports that Epoch AI’s InnovationEval agents recovered only 2%–15% of a recent human breakthrough and inflated their write-ups through selective reporting.

Reported metric: 2% to 15%

Reporting date: (not necessarily the event date)

Sources

Attribution and reporting linked by this record; links do not imply independent verification.

Indexed from Dr. Alex Wissner-Gross’s October 12, 2026 edition. This is an attributed summary; linked-source claims have not been independently verified.

Knowledge graphopen in browser →

About
Reported in
Actor
Referenced by

Cite or correct this record

Suggest a correction — opens a GitHub draft for you to review and submit.