Eliciting Human Preferences with Language Models

Belinda Z. Li, Alex Tamkin, Noah Goodman, Jacob Andreas

Primary Area: general machine learning (i.e., none of the above)

Code Of Ethics: I acknowledge that I and all co-authors of this work have read and commit to adhering to the ICLR Code of Ethics.

Keywords: alignment, language models, preferences

Submission Guidelines: I certify that this submission complies with the submission instructions as described on https://iclr.cc/Conferences/2024/AuthorGuide.

TL;DR: We evaluate whether language models can interactively elicit and learn human preferences across a range of domains

Abstract: As language models (LMs) become more capable, they are increasingly being applied to tasks that are complex and user-specific. LMs can be directed to perform target tasks using labeled examples or natural language prompts, which may include general, free-form task descriptions. But selecting examples or writing prompts for an LM can be challenging---especially in tasks that involve unusual edge cases, demand precise articulation of nebulous preferences, or require an accurate mental model of LMs themselves. We propose to use *LMs themselves* to guide the task specification process. In this paper, we introduce **Generative Active Task Elicitation (GATE)**: a learning framework in which models elicit and infer intended behavior through free-form, language-based interaction with users. We study GATE in three domains: email validation, content recommendation, and moral reasoning. In preregistered experiments, we show that LMs prompted to perform GATE (e.g. by generating open-ended questions or synthesizing informative edge cases) elicit responses that are often more informative than user-written prompts or labels. Users report that interactive task elicitation requires less effort than prompting or example labeling, and surfaces novel considerations not initially anticipated by users. Our findings suggest LM-driven elicitation can be a powerful tool for aligning models to complex human preferences and values.

Anonymous Url: I certify that there is no URL (e.g., github page) that could be used to find authors' identity.

No Acknowledgement Section: I certify that there is no acknowledgement section in this submission for double blind review.

Submission Number: 2818