Exploring Human Perceptions of AI Responses: Insights from a Mixed-Methods Study on Risk Mitigation in Generative Models

TL;DR

This study investigates how humans perceive AI responses, especially mitigation strategies, revealing factors influencing evaluations and introducing new metrics for assessing mitigation effectiveness.

Contribution

It provides novel insights into human perceptions of AI response mitigation and introduces new metrics for evaluating mitigation strategies in generative models.

Findings

Participants' native language and experience influence evaluations.

Minor grammar errors affect perceived response quality.

New metrics improve assessment of mitigation strategies.

Abstract

With the rapid uptake of generative AI, investigating human perceptions of generated responses has become crucial. A major challenge is their `aptitude' for hallucinating and generating harmful contents. Despite major efforts for implementing guardrails, human perceptions of these mitigation strategies are largely unknown. We conducted a mixed-method experiment for evaluating the responses of a mitigation strategy across multiple-dimensions: faithfulness, fairness, harm-removal capacity, and relevance. In a within-subject study design, 57 participants assessed the responses under two conditions: harmful response plus its mitigation and solely mitigated response. Results revealed that participants' native language, AI work experience, and annotation familiarity significantly influenced evaluations. Participants showed high sensitivity to linguistic and contextual attributes, penalizing minor grammar errors while rewarding preserved semantic contexts. This contrasts with how language is often treated in the quantitative evaluation of LLMs. We also introduced new metrics for training and evaluating mitigation strategies and insights for human-AI evaluation studies.