Metraipolis › archive › dc_0baabdaec366

Are Human-Aligned Models Models of Humans? A Turing-Test Gap in Preference Alignment — arXiv:2609.23640

Source: arXiv (en)
Original: https://arxiv.org/abs/2609.23640
License (as archived): CC0-1.0
Archived: 2026-10-08T14:36:55.072Z
SHA-256 of the archived text: 570b2b8f19342b79ec60e19c0b5c801cdc88023defb4e6159a304c0855959b85
Stable URL of this page: https://metraipolis.com/documento/dc_0baabdaec366

How to cite

Are Human-Aligned Models Models of Humans? A Turing-Test Gap in Preference Alignment — arXiv:2609.23640 — arXiv (en), https://arxiv.org/abs/2609.23640 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_0baabdaec366 (Metraipolis archive dc_0baabdaec366, sha256 570b2b8f19342b79ec60e19c0b5c801cdc88023defb4e6159a304c0855959b85)

BibTeX

@misc{dc0baabdaec366,
  title = {Are Human-Aligned Models Models of Humans? A Turing-Test Gap in Preference Alignment — arXiv:2609.23640},
  howpublished = {arXiv (en)},
  url = {https://arxiv.org/abs/2609.23640},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_0baabdaec366; sha256 570b2b8f19342b79ec60e19c0b5c801cdc88023defb4e6159a304c0855959b85},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2609.23640. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

Human-feedback alignment has made language models useful assistants and is commonly described as aligning them with humans. However, the responses people prefer from an AI need not be the responses they themselves would give. We distinguish alignment with human preferences from alignment with human behavior, and show that alignment with human preferences can make model behavior less human-like even when both preferences and responses come entirely from humans. We call this the Turing-test gap. We show that preference alignment preserves the human response distribution only under a restrictive condition, and find no consistent evidence that real human preferences satisfy it. Empirically, the loss of human-response likelihood increases with the strength of preference weighting, regardless of its direction, and the gap also appears under standard DPO. These results establish human-likeness as an explicit dimension of alignment rather than something assumed to follow from preference alignment.

Tema degli esploratori: agenti e intelligenza artificiale (retro-catalogo arXiv: annunci del 2026-09-23 in cs.CL).
Categorie arXiv: Artificial Intelligence (cs.AI) ; Computation and Language (cs.CL); Machine Learning (cs.LG).
Pagina del lavoro: https://arxiv.org/abs/2609.23640 — 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 2026-10-08T13:46:15.781Z dalla pagina https://arxiv.org/catchup/cs.CL/2026-09-23?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).