Metraipolis › archive › dc_7c8ddf90fd2d

Referential Uncertainty in Human--AI Collaboration — arXiv:2609.39518

Source: arXiv (en)
Original: https://arxiv.org/abs/2609.39518
License (as archived): CC0-1.0
Archived: 2026-10-07T12:42:49.960Z
SHA-256 of the archived text: d62bdb3076b0418a19ed379585578787240a029fe259966f952da7f45778b833
Stable URL of this page: https://metraipolis.com/documento/dc_7c8ddf90fd2d

How to cite

Referential Uncertainty in Human--AI Collaboration — arXiv:2609.39518 — arXiv (en), https://arxiv.org/abs/2609.39518 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_7c8ddf90fd2d (Metraipolis archive dc_7c8ddf90fd2d, sha256 d62bdb3076b0418a19ed379585578787240a029fe259966f952da7f45778b833)

BibTeX

@misc{dc7c8ddf90fd2d,
  title = {Referential Uncertainty in Human--AI Collaboration — arXiv:2609.39518},
  howpublished = {arXiv (en)},
  url = {https://arxiv.org/abs/2609.39518},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_7c8ddf90fd2d; sha256 d62bdb3076b0418a19ed379585578787240a029fe259966f952da7f45778b833},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2609.39518. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

Effective human-AI collaboration requires partners to establish references through interaction, which becomes fragile when descriptions are ambiguous, similar referents compete, or partners see different things. We study referential uncertainty - uncertainty over which candidate object a description refers to - in a collaborative puzzle task where a human Helper instructs an AI Worker to place pieces. The Worker must identify and communicate its uncertainty, and the Helper must recognize and act on it. We show that a separately elicited belief distribution over candidate pieces is better calibrated (ECE 0.15) and better discriminates correct from incorrect placements (AUROC 0.65) than raw action-token probabilities, which are severely overconfident (0.97 mean confidence, ECE 0.44). Across three frontier vision-language models (GPT-4.1, GPT-5, GPT-5.5), this elicited uncertainty rises predictably with instruction vagueness, but not with competing referents in context, even when those increase errors. The models seldom externalize it, asking for clarification on only 3.5-16.7% of turns. In a controlled human study (N=210), participants given only the Worker's default message accept 78% of wrong placements and cannot tell right from wrong (AUC 0.50). Precise descriptions and, especially, well-targeted hedges cut wrong-move acceptance to 36% while largely preserving correct-move acceptance, compensating for missing shared awareness such as not seeing the Worker's action. But this benefit depends on targeting: a deployable hedge derived from the model's own belief entropy inherits that signal's weakness and can do more harm than good. Externalized uncertainty helps a human partner only when it is accurately targeted.

Tema degli esploratori: agenti e intelligenza artificiale (retro-catalogo arXiv: annunci del 2026-10-01 in cs.AI).
Categorie arXiv: Artificial Intelligence (cs.AI) ; Human-Computer Interaction (cs.HC).
Pagina del lavoro: https://arxiv.org/abs/2609.39518 — 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 2026-10-07T12:14:13.770Z dalla pagina https://arxiv.org/catchup/cs.AI/2026-10-01?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).