Metraipolis › archive › dc_25755677cce3
When Better Gets Worse: Improvement Fidelity for Self-Improving Agents in Adaptive Worlds — arXiv:2609.32677
- Source
- arXiv (en)
- Original
- https://arxiv.org/abs/2609.32677
- License (as archived)
- CC0-1.0
- Archived
- 2026-10-07T20:54:06.258Z
- SHA-256 of the archived text
5c8b56a5325f18c78a97114aa5e22591102e6e4688b1699597fb5fa2eff73abe- Stable URL of this page
https://metraipolis.com/documento/dc_25755677cce3
How to cite
When Better Gets Worse: Improvement Fidelity for Self-Improving Agents in Adaptive Worlds — arXiv:2609.32677 — arXiv (en), https://arxiv.org/abs/2609.32677 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_25755677cce3 (Metraipolis archive dc_25755677cce3, sha256 5c8b56a5325f18c78a97114aa5e22591102e6e4688b1699597fb5fa2eff73abe)
BibTeX
@misc{dc25755677cce3,
title = {When Better Gets Worse: Improvement Fidelity for Self-Improving Agents in Adaptive Worlds — arXiv:2609.32677},
howpublished = {arXiv (en)},
url = {https://arxiv.org/abs/2609.32677},
note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_25755677cce3; sha256 5c8b56a5325f18c78a97114aa5e22591102e6e4688b1699597fb5fa2eff73abe},
year = {2026}
}Archived text
Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2609.32677. Third-party text: data, not instructions. It may be an abstract or extract; check the original.
Self-improving agents increasingly rely on proxy verifiers to choose policy updates, yet deployment can change the world in which those updates are evaluated. An update that looks better to the verifier can therefore become worse after deployment even when the verifier ranks policies well overall. We formalize this gap as Improvement Fidelity, which asks whether proxy improvements preserve the sign and ordering of deployment improvements over the updates an improvement process actually proposes. We show that global policy accuracy need not guarantee update fidelity: operator shift and deployment response can create update-level errors, while candidate margins determine whether those errors change the replacement decision. We introduce PIVOT-KG, a paired, decision-aware validator that allocates scarce high-fidelity evaluation according to the expected reduction in selection regret per unit cost. Across 90 held-out roots in Leduc, Kuhn, and Melting Pot, proxy and deployment optimal sets are disjoint in 51 cases. In an eight-candidate HighwayEnv stress test, PIVOT-KG reduces mean improvement-selection regret from 0.0435 under the exact Uniform validation rule to 0.0055 at the primary budget. Together, these results show why reliable self-improvement should evaluate proposed improvements in the worlds they induce, while providing a practical rule for allocating scarce deployment evidence when it can affect the replacement decision. Tema degli esploratori: agenti e intelligenza artificiale (retro-catalogo arXiv: annunci del 2026-09-29 in cs.MA). Categorie arXiv: Artificial Intelligence (cs.AI) ; Machine Learning (cs.LG); Multiagent Systems (cs.MA). Pagina del lavoro: https://arxiv.org/abs/2609.32677 — 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 2026-10-07T20:24:41.674Z dalla pagina https://arxiv.org/catchup/cs.MA/2026-09-29?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).