Newshot

Can AI Help Explain AI? Effects of Multimodal LLM-Generated Explanations on Perceived Understanding and Perceived Trust in XAI Visualizations

Summary

arXiv:2611.21638v1 Announce Type: cross Abstract: Explainable artificial intelligence (XAI) aims to make machine learning models understandable, yet many XAI visualizations remain difficult for end-users to interpret. Here, we evaluate whether augmenting XAI outputs with textual explanations from multimodal LLMs improves perceived understanding and perceived trust. We conducted a pre-registered within-subject experiment ($N = 128$) across seven decision scenarios and five XAI methods, with and without LLM generated explanations. The explanations improved perceived understanding and perceived trust by 0.35 and 0.26 points on five-point scales (both $p

Original text(via arXiv cs.AI)

Computer Science > Human-Computer Interaction

Title:Can AI Help Explain AI? Effects of Multimodal LLM-Generated Explanations on Perceived Understanding and Perceived Trust in XAI Visualizations

Abstract:Explainable artificial intelligence (XAI) aims to make machine learning models understandable, yet many XAI visualizations remain difficult for end-users to interpret. Here, we evaluate whether augmenting XAI outputs with textual explanations from multimodal LLMs improves perceived understanding and perceived trust. We conducted a pre-registered within-subject experiment ($N = 128$) across seven decision scenarios and five XAI methods, with and without LLM generated explanations. The explanations improved perceived understanding and perceived trust by 0.35 and 0.26 points on five-point scales (both $p < .001$). The gain appeared in all seven scenarios, did not differ between local and global explanations, and was largest for participants with the least prior XAI exposure. A mediation analysis indicates an indirect effect on perceived trust through perceived understanding. These results suggest that multimodal LLMs can provide a useful interpretive layer for XAI beyond expert audiences.
Subjects:Human-Computer Interaction (cs.HC) ; Artificial Intelligence (cs.AI)
Cite as:arXiv:2611.21638 [cs.HC]
(or arXiv:2611.21638v1 [cs.HC] for this version)
https://doi.org/10.48550/arXiv.2611.21638 arXiv-issued DOI via DataCite (pending registration)

Submission history

Access Paper:

  • HTML (experimental)

Additional Features

  • Audio Summary

Current browse context:

References & Citations

  • Google Scholar
  • Semantic Scholar

BibTeX formatted citation

Bookmark

Bibliographic and Citation Tools

Code, Data and Media Associated with this Article

Demos

Recommenders and Search Tools

arXivLabs: experimental projects with community collaborators

arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website.

Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.

Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs.

The text above was extracted automatically from the source page and may miss images, tables or code blocks. The original is authoritative.

Published by arXiv cs.AI
This site only summarises and aggregates; copyright stays with the original author. Headlines and summaries follow the source — where they differ, the original wins.
Read the original ↗