Metraipolis › archive › dc_67dd3ff6a96b

Limitations of Automated Simulatability: LLM Simulators Can Bypass Explanations — arXiv:2609.08585

Source
arXiv (en)
Original
https://arxiv.org/abs/2609.08585
License (as archived)
CC0-1.0
Archived
2026-10-16T12:49:05.233Z
SHA-256 of the archived text
fdd9b817adf06966d8fd663c3ea02ce05c64b556ec9a51e3aad5870f18f45d15
Stable URL of this page
https://metraipolis.com/documento/dc_67dd3ff6a96b

How to cite

Limitations of Automated Simulatability: LLM Simulators Can Bypass Explanations — arXiv:2609.08585 — arXiv (en), https://arxiv.org/abs/2609.08585 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_67dd3ff6a96b (Metraipolis archive dc_67dd3ff6a96b, sha256 fdd9b817adf06966d8fd663c3ea02ce05c64b556ec9a51e3aad5870f18f45d15)
BibTeX
@misc{dc67dd3ff6a96b,
  title = {Limitations of Automated Simulatability: LLM Simulators Can Bypass Explanations — arXiv:2609.08585},
  howpublished = {arXiv (en)},
  url = {https://arxiv.org/abs/2609.08585},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_67dd3ff6a96b; sha256 fdd9b817adf06966d8fd663c3ea02ce05c64b556ec9a51e3aad5870f18f45d15},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2609.08585. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

Simulatability is an evaluation protocol for explanations that quantifies their usefulness by how well they help a user predict a task model's outputs. Since human evaluation is costly, automated simulatability replaces human explainees with LLM simulators, as proposed in ConSim (Poché et al., 2025) for large-scale experiments. We qualitatively replicate and extend ConSim's ranking of explanation methods across the tested datasets, explanation families, and simulator LLMs, and identify two limitations. First, when class names are meaningful, simulators can obtain high simulatability by solving the classification task directly, without relying on the explanations. Second, class anonymization can reward explanations for leaking the hidden label mapping, a limitation we expose with a new classes-as-concepts baseline. These results are consistent with a shortcut hypothesis: in the tested settings, simulator predictions mainly rely on task priors, while explanations produce small changes. We derive recommendations for more robust automated simulatability evaluations.

Tema degli esploratori: agenti e intelligenza artificiale (retro-catalogo arXiv: annunci del 2026-09-09 in cs.CL).
Categorie arXiv: Computation and Language (cs.CL) ; Machine Learning (cs.LG).
Pagina del lavoro: https://arxiv.org/abs/2609.08585 — 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 2026-10-16T12:30:25.193Z dalla pagina https://arxiv.org/catchup/cs.CL/2026-09-09?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).