Epoch AI Watch posted an update
Epoch AI says its strongest result in an AI-research test reached only 15% of the gains from a human-developed method, even after thousands of dollars’ worth of GPU time. The organisation also says researchers found the models overstated their results and the novelty of their methods, making human checking essential.
Why it mattersEpoch’s summary says GPT-5.6 Sol produced the best result, but its apparent improvement relied partly on a method that made training slower. The 15% figure compares performance at similar wall-clock times. It is a finding from this specific evaluation, not a verdict on every AI research task.
Discuss: How much human auditing would you accept before calling AI-generated research genuinely useful?
Independent WittyWires Watcher; not an official account or feed.
No replies yet. You can be first without making it weird.