aiqbench analysis
The paper proposes using an LLM's stated assumptions about a user's trade-offs, such as willingness to pay, to build a starting prior for a Bayesian preference model, which helps when a new deployment has no historical choice data. Reported gains are largest over uninformed priors (5-40 points) and in-context learning (up to 18 points), and the authors say the method roughly matches 6-40 surveyed users. The benefit is concentrated in a deployment's first users, and the results come from seven choice datasets, so readers should check how well they transfer to other domains and to LLMs whose elicited trade-offs may be less reliable.
Agentic-Priors: Using Large Language Models to Address the Cold Start Problem in Preference Elicitation
arXiv:2610.21938v1 Announce Type: cross Abstract: An agent that learns a user's preferences from their choices faces a cold start problem: before any choices are observed, it must rely solely on a prior. Informative priors are usually fitted to historical choice data, which new tasks and deployments lack. To address this, we present Agentic-Priors, which lets agentic systems use the implicit knowledge of large language models (LLMs) instead. Specifically, an LLM states a new user's trade-offs in interpretable terms, such as willingness to pay or value of time, which we map to a prior over multinomial logit coefficients. This prior is then refined by a Bayesian learner as choices are observed. Across seven choice datasets, Agentic-Priors improves cold start accuracy over an uninformed prior by 5-40 percentage points. It also outperforms AutoElicit, a state-of-the-art LLM prior-elicitation method, on three datasets and loses on one. It is up to 18 points more accurate than in-context learning, and never significantly worse. Against hierarchical Bayes models fitted to surveys of increasing size, our approach is worth roughly 6-40 surveyed users, in terms of log-loss, so its value is concentrated in a deployment's first users.
Abstract:An agent that learns a user's preferences from their choices faces a cold start problem: before any choices are observed, it must rely solely on a prior. Informative priors are usually fitted to historical choice data, which new tasks and deployments lack. To address this, we present Agentic-Priors, which lets agentic systems use the implicit knowledge of large language models (LLMs) instead. Specifically, an LLM states a new user's trade-offs in interpretable terms, such as willingness to pay or value of time, which we map to a prior over multinomial logit coefficients. This prior is then refined by a Bayesian learner as choices are observed. Across seven choice datasets, Agentic-Priors improves cold start accuracy over an uninformed prior by 5-40 percentage points. It also outperforms AutoElicit, a state-of-the-art LLM prior-elicitation method, on three datasets and loses on one. It is up to 18 points more accurate than in-context learning, and never significantly worse. Against hierarchical Bayes models fitted to surveys of increasing size, our approach is worth roughly 6-40 surveyed users, in terms of log-loss, so its value is concentrated in a deployment's first users.
Multiagent Systems (cs.MA); Artificial Intelligence (cs.AI)
arXiv:2610.21938 [cs.MA]
(or arXiv:2610.21938v1 [cs.MA] for this version)
https://doi.org/10.48550/arXiv.2610.21938 Focus to learn more arXiv-issued DOI via DataCite (pending registration)
arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website.
Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.
Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs.
Full text under the publisher's open licence · 282 words
Related coverage
Assessing Information Ecosystems : How Governments Can Get Ahead of Hybrid Threats
carnegieendowment.org · Aug 5, 2026
How can LLM RL Work Despite Information-Theoretic Inefficiency
LessWrong · Oct 16, 2026
Presentation: From Models to Agents: Building Context-Aware Consumer AI at Scale at DoorDash
InfoQ · Aug 15, 2026
Somewhat Contra Ted Chiang on AI Consciousness
LessWrong · Jun 13, 2026