# UIUC paper says AI novelty checks reject ideas humans later publish

> Original title: New University of Illinois paper finds that when AI decides if a research idea is new enough to work on, it often says no to ideas that h...

- Source: Rohan Paul
- Published: 2026-10-31T20:00:57.000Z
- Bad Influence: https://badinfluence.news/items/tpqx8fl0zimnv998tu7g4lzh3
- Original: https://x.com/rohanpaul_ai/status/2116621467219816915

## Summary

A University of Illinois paper finds that AI novelty checks for generated research ideas often reject proposals that human scientists later publish. Its ClaimNov system splits each proposal into claims and checks them one by one against prior papers. ClaimNov caught 97% of disguised copies and rejected only 1.5% of later-published ideas, versus 96% for a deployed whole-proposal check.

## Post

New University of Illinois paper finds that when AI decides if a research idea is new enough to work on, it often says no to ideas that human scientists later publish.

AI scientist pipelines run a novelty check to decide which generated ideas get compute. These checks judge a whole proposal at once, though most real papers build on some earlier contributions.

They built ClaimNov, which splits a proposal into a few claims, matches each against claims from prior papers, and gives a judge the claim-by-claim evidence.

A deployed check caught all 380 disguised copies but also rejected 96% of ideas researchers later published. ClaimNov caught 97% of copies while rejecting only 1.5% of those ideas.

If your pipeline filters ideas for novelty, score claims separately and keep proposals that build on 1 known result instead of discarding them.

![](https://badinfluence.news/api/img-proxy?u=https%3A%2F%2Fpbs.twimg.com%2Fmedia%2FHV--lwVbUAAW4mS.jpg&mode=thumb&exp=1793664000&sig=5ab4a9fff6dac88e)