analiza aiqbench
This research identifies a fundamental gap in Large Vision-Language Models regarding their ability to model "common ground" during interactive referential communication. The study demonstrates that LVLMs struggle to generate and resolve referring expressions effectively when collaborating with humans or other AIs on tasks involving non-lexicalized objects, limiting the smoothness of human-AI partnership. By releasing a corpus of 356 dialogues alongside analysis tools, this work provides essential data for understanding the specific constraints hindering naturalistic AI collaboration.
LVLMs and Humans Ground Differently in Referential Communication
arXiv:2601.19792v4 Announce Type: replace-cross Abstract: For generative AI agents to partner effectively with human users, the ability to accurately predict human intent is critical. But this ability to collaborate remains limited by a critical deficit: an inability to model common ground. We present a referential communication experiment with a factorial design involving director-matcher pairs (human-human, human-AI, AI-human, and AI-AI) that interact with multiple turns in repeated rounds to match pictures of objects not associated with any obvious lexicalized labels. We show that LVLMs cannot interactively generate and resolve referring expressions in a way that enables smooth communication, a crucial skill that underlies human language use. We release our corpus of 356 dialogues (89 pairs over 4 rounds each) along with the online pipeline for data collection and the tools for analyzing accuracy, efficiency, and lexical overlap.
Abstract:For generative AI agents to partner effectively with human users, the ability to accurately predict human intent is critical. But this ability to collaborate remains limited by a critical deficit: an inability to model common ground. We present a referential communication experiment with a factorial design involving director-matcher pairs (human-human, human-AI, AI-human, and AI-AI) that interact with multiple turns in repeated rounds to match pictures of objects not associated with any obvious lexicalized labels. We show that LVLMs cannot interactively generate and resolve referring expressions in a way that enables smooth communication, a crucial skill that underlies human language use. We release our corpus of 356 dialogues (89 pairs over 4 rounds each) along with the online pipeline for data collection and the tools for analyzing accuracy, efficiency, and lexical overlap.
Computation and Language (cs.CL); Artificial Intelligence (cs.AI); Human-Computer Interaction (cs.HC)
arXiv:2601.19792 [cs.CL]
(or arXiv:2601.19792v4 [cs.CL] for this version)
Text integral sub licența deschisă a publicației · 149 cuvinte
Subiecte conexe
Finding heterogeneous agent swarms in the wild
LessWrong · 16 sept. 2026
OpenMAS-GCom. A Diagnostic Benchmark for Graph-enhanced Multi-Agent Systems
arXiv · 21 sept. 2026
Podcast: AI, ML, and Data Engineering InfoQ Trends Report 2026
InfoQ · 30 oct. 2026
Links #3: 2026/06 Part 1
LessWrong · 15 iun. 2026