Can AI Extract Antecedent Factors of Human Trust in AI? An Application of Information Extraction for Scientific Literature in Behavioural and Computer Sciences

TL;DR

This paper develops a new annotated dataset and benchmarks methods for extracting factors influencing human trust in AI from scientific literature, highlighting the need for supervised learning approaches.

Contribution

The paper introduces the first annotated dataset for trust factors in AI from scientific texts and evaluates LLM-guided annotation and extraction methods.

Findings

Supervised learning outperforms prompt-based LLMs for this task.

Created the first annotated dataset in this domain.

Benchmark results show current LLMs are insufficient for accurate extraction.

Abstract

Information extraction from the scientific literature is one of the main techniques to transform unstructured knowledge hidden in the text into structured data which can then be used for decision-making in down-stream tasks. One such area is Trust in AI, where factors contributing to human trust in artificial intelligence applications are studied. The relationships of these factors with human trust in such applications are complex. We hence explore this space from the lens of information extraction where, with the input of domain experts, we carefully design annotation guidelines, create the first annotated English dataset in this domain, investigate an LLM-guided annotation, and benchmark it with state-of-the-art methods using large language models in named entity and relation extraction. Our results indicate that this problem requires supervised learning which may not be currently feasible with prompt-based LLMs.