FuguReport

Evidence-Informed LLM Beliefs for Continual Scientific Discovery

Authors Dhruv Agarwal, Reece Adamson, Andrew McCallum, Peter Clark, Ashish Sabharwal, Bodhisattwa Prasad Majumder
Affiliations University of Massachusetts Amherst / Allen Institute for AI
Categories Method / Belief Revision / Filtering and updating beliefs, Method / Retrieval-Augmented Generation / Embedding-based search-enhanced generation, Application / Scientific Discovery / Prioritizing surprising hypotheses
License CC BY 4.0

Abstract Overview

This paper studies continual scientific discovery with LLMs and argues that prior work, especially AutoDiscovery, measures surprise against a static prior even though beliefs should evolve as evidence accumulates. The authors introduce evidence-informed LLM beliefs, where the prior for a new hypothesis is updated using outcomes from previously tested hypotheses, yielding a non-stationary notion of surprisal. They compare several in-context belief-update mechanisms and find that embedding-based retrieval over prior discoveries best aligns prior beliefs with eventual posteriors. They also modify the search process with belief-update filtering and diversity maximization so that the system avoids redundant hypotheses and explores regions that remain surprising under updated beliefs.

Novelty

The main novelty is reframing LLM-driven scientific discovery around non-stationary surprisal, computed from priors that are continually updated with prior evidence rather than treated as fixed. The work also introduces an evidence-informed search procedure that couples belief-update filtering with diversity maximization to make search responsive to those updated beliefs.

Results

Embedding-based top-k retrieval performed best among the tested belief-update mechanisms, identifying about 37.5% of static surprisals as spurious. Across five discovery domains, the full evidence-informed search method increased accumulated non-stationary surprisal by 30.62% over standard static-reward search, demonstrating gains beyond online deduplication alone.

Key Points

  1. Static surprisal can overcount discovery progress because hypotheses implied by earlier findings may still appear surprising when priors are not updated.
  2. Among the tested context representations, top-k embedding-based retrieval was the most effective way to construct evidence-informed priors while being more efficient than full-context conditioning.
  3. The search improvements come from explicit hypothesis-selection changes—belief-update filtering, diversity maximization, and online deduplication—rather than simply swapping the reward from static to non-stationary surprisal.

References

This page was created using generative AI such as GPT-5, Claude Opus 4, Gemini 3, Gemini 3.1 Flash Image, and their higher-end successor versions. No guarantee can be made regarding its contents.