Mitigating Misleadingness in LLM-Generated Natural Language Explanations for Recommender Systems: Ensuring Broad Truthfulness Through Factuality and Faithfulness
Year: 2025
Type: article
Abstract: Large Language Models (LLMs) are increasingly used to generate natural language explanations in recommender systems, offering the potential for adaptive, personalized, and interactive communication.However, their ability to produce plausible yet inaccurate justifications poses a risk of misleading users at scale.This paper examines the risk of misleading LLM-generated explanations, focusing on the interplay between truthfulness and persuasiveness.We propose a clear terminology for categorizing misleading explanations, differentiating between factual errors and unfaithful representations of the system's workings.Drawing on research from machine learning, human-computer interaction, and persuasive communication, we identify key challenges in evaluating and mitigating misleading explanations.We outline directions for future research, including the development of robust evaluation metrics and methods for enhancing the factuality and faithfulness of LLM-generated explanations in recommender systems.
Author: Ulysse Maes
Institutions: Vrije Universiteit Brussel, IMEC
Language: English
Subfield: Artificial Intelligence
Field: Computer Science
Domain: Physical Sciences
Open Access status: green
OpenAlex ID: W7112902589