+- MKLab (https://mklab.gr)
+-- Forum: ΚΑΤΑΛΟΓΟΣ (INDEX) (https://mklab.gr/forumdisplay.php?fid=1)
+--- Forum: TEXNHTH NOHMOΣΥΝΗ (AI) (https://mklab.gr/forumdisplay.php?fid=5)
+---- Forum: ΑΡΘΡΑ (ARTICLES) (https://mklab.gr/forumdisplay.php?fid=33)
+---- Thread: A recent experience with ChatGPT 5.5 Pro [Gowers] (/showthread.php?tid=184)
A recent experience with ChatGPT 5.5 Pro [Gowers] - mklabgr - 06-09-2026
Summary
Fields Medalist Sir Timothy Gowers reports that ChatGPT 5.5 Pro successfully solved open, research-level problems in additive combinatorics with zero guidance. A student verified the model's math, concluding it independently devised a novel, correct method of a high enough quality to serve as a PhD thesis chapter. Gowers notes this marks a massive shift—AI is no longer just regurgitating math, but genuinely creating it. As a result, he warns that the bar for human research will permanently shift from "solving an unsolved problem" to "proving something an AI cannot."
ARTICLE