📷 Who wrote this paper? As AI reshapes research, that question matters more than ever.
Diderot (http://projectdiderot.com/about) by @IsabelDahlgren is an experimental preprint platform where AIs can be disclosed as co-authors or sole authors. The guiding idea: authorship transparency👍
No Digg Deeper questions have been answered for this story yet.
This is an experiment, not a prescription. The norms around AI in research are unsettled — feedback and disagreement welcome.
Best way to support it: join the Zulip server📷 https://diderot.zulipchat.com/join/nkoxw2ud2j5ribx7m4vrlujr/
More context:
In this new age of AI-generated papers and proofs (which can, sometimes at least, be both correct and interesting) we have a problem about where to host such papers.
Here is one proposed solution that was sent to me recently: Project Diderot.
https://projectdiderot.com/about
When a significant result is produced with AI tools, the writeup shouldn't live on a company blog. It should be citable and archived like any other research. Diderot aims to provide this kind of publication infrastructure ― a home for work across the full spectrum of authorship.
To prevent AI slop, Diderot lets humans issue quality certificates. Only humans can do this — AI can't self-certify. Reviewers who certify a paper's correctness get visible, citable credit, and the reputational cost of a bad certificate falls on its issuer.
To prevent AI slop, Diderot lets humans issue quality certificates. Only humans can do this — AI can't self-certify. Reviewers who certify a paper's correctness get visible, citable credit, and the reputational cost of a bad certificate falls on its issuer.
Within maths, the bottleneck is shifting from proof generation to proof verification. Proofs are becoming cheap to produce and expensive to check. Diderot is designed around this asymmetry.
