Limits to black-box amplification in QMA, with the key step written by GPT-5
A quantum complexity paper proving black-box error amplification for QMA cannot push completeness closer to certainty than doubly exponentially, in which the pivotal technical step was produced by GPT-5 in about half an hour.
- Model
- GPT-5 Thinking
- Field
- Computer science
- Date
- 2025-09-25
- Human collaborators
- Scott Aaronson, Freek Witteveen
Sources
Original work
Announcement
What was found
The authors were stuck on how the largest eigenvalue of a matrix behaved as a parameter varied. GPT-5 proposed reformulating the quantity so that a complex approximation-theory bound applied, and that reformulation became the technical core of the oracle separation. Aaronson wrote that the step would have cost him or a graduate student a couple of weeks and that it was the first substantive AI contribution to a paper of his. The result makes his 2008 oracle separation quantitative and shows recent amplification results are optimal for black-box procedures.
Novelty check
The theorem extends Aaronson's 2008 QMA oracle separation and a 2025 result of Jeffery and Witteveen; no prior bound of this form on black-box amplification appears in the literature. The AI contribution is documented in the paper's acknowledgements and on Aaronson's own blog, not inferred from press coverage. That matters, because most "AI proved a theorem" stories are not.
Caveats and known objections
An arXiv preprint. The contribution is one lemma inside a human-conceived and human-written paper; Aaronson is explicit that the model did not pose the problem or design the argument and that he could have done the step himself given time. Autonomy is ai-assisted, which stays the right grade even though the step was pivotal.
Nobody outside the lab has checked this yet.
Reading the primary source closely enough to say whether it supports the claim counts as a check, and you are credited on the entry.
Or on GitHub: submit a check challenge the grade send a correction or send a pull request
Flag this for triage
Signals order the review queue and nothing else. They are never published, and they never move a grade: that takes a citation.
Entry history (1 event)
- AddedEntered the registry graded Author verified and AI-assisted.
Entries are never deleted. A grade that does not hold up is downgraded on the record, with the reason beside it.
Graded author verified for verification and ai-assisted for autonomy. What these mean.
Cite this entry
whataifound.org. (2025). Limits to black-box amplification in QMA, with the key step written by GPT-5. whataifound.org: A Registry of AI Scientific and Mathematical Discoveries. https://whataifound.org/finding/2025-09-25-qma-amplification-limitsBibTeX
@misc{whataifound-utaustin-2025-limits,
title = {Limits to black-box amplification in QMA, with the key step written by GPT-5},
author = {{whataifound.org}},
year = {2025},
howpublished = {whataifound.org: A Registry of AI Scientific and Mathematical Discoveries},
note = {Result by UT Austin / CWI Amsterdam. Verification: Author verified. Autonomy: AI-assisted.},
url = {https://whataifound.org/finding/2025-09-25-qma-amplification-limits}
}