Preference elicitation systems aim to learn what a user wants from their choices, but they face a cold start problem: before any choices are observed, the agent must rely solely on a prior. In practice, informative priors are usually fitted to historical choice data, which new users may not have. A new arXiv paper, "Agentic-Priors: Using Large Language Models to Address the Cold Start Problem in Preference Elicitation," proposes an alternative.

Instead of historical data, the paper suggests using large language models to generate priors. This approach, termed agentic priors, would allow an agent to draw on general knowledge encoded in an LLM to make initial guesses about user preferences. This could help systems personalize recommendations or decisions from the very first interaction.

The abstract is truncated, so details on implementation and evaluation are not available from the source. The paper is listed as new on arXiv under identifier 2610.21938v1.