Eric Topol and Derya Unutmaz clash over extrapolating GPT-4 study findings to frontier models
DeryaTR_ · x · 2026-10-27
Eric Topol and Derya Unutmaz traded blows over a study questioning whether frontier-model use leads to sustained upskilling.
- Topol: the study's fundamental question — whether current frontier models produce sustained upskilling — remains unanswered; the GPT-4 comparability complaint dodges it
- Unutmaz: the burden of proof isn't on him; you cannot extrapolate findings from an obsolete model like GPT-4 to GPT-6, and this reflects a scientific publication process that needs a complete overhaul
The core dispute: whether empirical results from outdated models can support general claims about AI, and whether peer review should filter out such studies.
Related event: Topol vs Derya: Can GPT-4 Skill-Loss Findings Generalize to Newer Models?(5 posts)→
More from AGI Musings
- OpenAI says its models produced a Yang-Mills existence proof, formalized in Lean — exophades · 2026-10-27
- Greg Brockman announces AI-assisted discoveries in math, CS, and quantum field theory — gdb · 2026-10-27
- AI is simultaneously flooding journals with slop and cleaning up human slop — _akpiper · 2026-10-27
- EMNLP researcher worries LLM-polished talks are erasing individual nuance — lucy3_li · 2026-10-27
- Paul Graham: Not Reading or Writing Will Become as Ubiquitous as Smoking Once Was — IanArawjo · 2026-10-27
- 2012 whole-cell model paper shows we can scalably measure only ~5 of the processes needed to simulate biology — niloofar_mire · 2026-10-27