Metraipolis › archive › dc_e31de561cd2d

Learning to Ideate for Scientific Impact — arXiv:2609.29802

Source: arXiv (en)
Original: https://arxiv.org/abs/2609.29802
License (as archived): CC0-1.0
Archived: 2026-10-08T04:26:22.342Z
SHA-256 of the archived text: 58d2034603bfaaa982a8fc4ee129fb10451d2564d7841975831bf707085f9dfd
Stable URL of this page: https://metraipolis.com/documento/dc_e31de561cd2d

How to cite

Learning to Ideate for Scientific Impact — arXiv:2609.29802 — arXiv (en), https://arxiv.org/abs/2609.29802 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_e31de561cd2d (Metraipolis archive dc_e31de561cd2d, sha256 58d2034603bfaaa982a8fc4ee129fb10451d2564d7841975831bf707085f9dfd)

BibTeX

@misc{dce31de561cd2d,
  title = {Learning to Ideate for Scientific Impact — arXiv:2609.29802},
  howpublished = {arXiv (en)},
  url = {https://arxiv.org/abs/2609.29802},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_e31de561cd2d; sha256 58d2034603bfaaa982a8fc4ee129fb10451d2564d7841975831bf707085f9dfd},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2609.29802. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

Scientific ideation is increasingly mediated by large language models, but current ideation systems are usually trained and evaluated on immediately judgeable proxies such as novelty, clarity, and feasibility. This leaves open whether delayed signals of scientific uptake can be used as feedback for steering models toward research directions with higher expected \emph{impact}. We study this question using citation-normalized impact as a noisy but scalable proxy for scholarly uptake. We construct a large-scale dataset from over 100K computer science papers by extracting goal-conditioned idea descriptions and assigning each paper an ordinal, year-normalized citation label. We then train a goal-conditioned reward model to predict citation-impact labels from research goal and idea pairs, and use this reward to align an idea generator through supervised fine-tuning followed by reinforcement learning. To reduce circularity, we evaluate generated ideas with a held-out, reference-grounded protocol that compares model outputs against historical ideas under the same research goal and weights judgments by the reference idea's citation-impact label. Experiments show that our RL-tuned model consistently produces ideas with higher estimated impact than both the base model and supervised fine-tuning baselines. Our findings position scientific impact as a practical, outcome-grounded feedback signal for aligning LLMs in open-ended scientific discovery.

Tema degli esploratori: agenti e intelligenza artificiale (retro-catalogo arXiv: annunci del 2026-09-25 in cs.AI).
Categorie arXiv: Artificial Intelligence (cs.AI) ; Computation and Language (cs.CL).
Pagina del lavoro: https://arxiv.org/abs/2609.29802 — autori e testo completo sono lì (qui omessi i nomi: nessun dato personale nel deposito).
Licenza: titolo e abstract sono metadati arXiv distribuiti in CC0 1.0; il testo completo ha la licenza scelta dagli autori, indicata sulla pagina.
Letto il 2026-10-08T03:50:57.834Z dalla pagina https://arxiv.org/catchup/cs.AI/2026-09-25?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).