How Do LLMs Encode Scientific Quality? An Empirical Study Using Monosemantic Features from Sparse Autoencoders

TL;DR

This study investigates how large language models encode scientific quality by extracting monosemantic features using sparse autoencoders, revealing multiple dimensions and key aspects of research quality representation.

Contribution

First empirical analysis of LLM internal representations of scientific quality using monosemantic features derived from autoencoders, linking features to research impact indicators.

Findings

LLMs encode features related to research methodologies and publication types.

Features associated with high-impact fields and scientific jargon are prominent.

Identified four key feature types that capture aspects of research quality.

Abstract

In recent years, there has been a growing use of generative AI, and large language models (LLMs) in particular, to support both the assessment and generation of scientific work. Although some studies have shown that LLMs can, to a certain extent, evaluate research according to perceived quality, our understanding of the internal mechanisms that enable this capability remains limited. This paper presents the first study that investigates how LLMs encode the concept of scientific quality through relevant monosemantic features extracted using sparse autoencoders. We derive such features under different experimental settings and assess their ability to serve as predictors across three tasks related to research quality: predicting citation count, journal SJR, and journal h-index. The results indicate that LLMs encode features associated with multiple dimensions of scientific quality. In particular, we identify four recurring types of features that capture key aspects of how research quality is represented: 1) features reflecting research methodologies; 2) features related to publication type, with literature reviews typically exhibiting higher impact; 3) features associated with high-impact research fields and technologies; and 4) features corresponding to specific scientific jargons. These findings represent an important step toward understanding how LLMs encapsulate concepts related to research quality.