Metraipolis › archive › dc_e16889133fa2

Science sandboxes measure the scientific capability of AI agents — arXiv:2608.30165

Source
arXiv (en)
Original
https://arxiv.org/abs/2608.30165
License (as archived)
CC0-1.0
Archived
2026-10-17T07:52:28.310Z
SHA-256 of the archived text
09ca215873ca185bb9296457ef497b82642a3a03013b16313f3c80bc90fe24b3
Stable URL of this page
https://metraipolis.com/documento/dc_e16889133fa2

How to cite

Science sandboxes measure the scientific capability of AI agents — arXiv:2608.30165 — arXiv (en), https://arxiv.org/abs/2608.30165 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_e16889133fa2 (Metraipolis archive dc_e16889133fa2, sha256 09ca215873ca185bb9296457ef497b82642a3a03013b16313f3c80bc90fe24b3)
BibTeX
@misc{dce16889133fa2,
  title = {Science sandboxes measure the scientific capability of AI agents — arXiv:2608.30165},
  howpublished = {arXiv (en)},
  url = {https://arxiv.org/abs/2608.30165},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_e16889133fa2; sha256 09ca215873ca185bb9296457ef497b82642a3a03013b16313f3c80bc90fe24b3},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2608.30165. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

Scientific progress depends not only on finding solutions, but on learning the rules that explain why they work and using that understanding to design better experiments. We introduce science sandboxes, a framework for studying this capability in AI agents through repeated cycles of experimentation, feedback, and hypothesis revision. Science sandboxes invite an agent to query the natural world in different ways, ranging from "wet" physical experiments, to "damp" predictive models trained on empirical data, to "dry" invented rules. By establishing a common experimental loop and a protocol for evaluating agents within it, science sandboxes allow assessment of both quantitative performance on specific metrics and qualitative scientific reasoning, across a spectrum of empirical verifiability. Here, we instantiate this framework in two biological settings, models of regulatory genomics and protein fitness prediction, and examine the capabilities of frontier agents. Across these settings, we could see when agents successfully optimized a quantitative metric without understanding the rules underlying the system. In particular, their scientific reasoning deteriorated when they encountered systems whose rules fell outside familiar biological priors. By highlighting such failure modes, science sandboxes make the frontier of scientific capability measurable and provide a controlled setting in which to study and ultimately expand it.

Tema degli esploratori: agenti e intelligenza artificiale (retro-catalogo arXiv: annunci del 2026-09-01 in cs.AI).
Categorie arXiv: Quantitative Methods (q-bio.QM) ; Artificial Intelligence (cs.AI).
Pagina del lavoro: https://arxiv.org/abs/2608.30165 — 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 2026-10-17T07:28:26.834Z dalla pagina https://arxiv.org/catchup/cs.AI/2026-09-01?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).