Addressing AI Vulnerabilities Through Human-Centered Approaches and Risk Frameworks
Sarvesh Sawant, Aasish Bhanu, Beau G. Schelble, Kapil Chalil Madathil
Advances in Human-AI Collaboration, John Wiley and Sons, pp. 325-339 (2026)
Overview
While AI offers various practical advantages, it is not without risks. AI can make mistakes when it is trained on incomplete or biased information. It may misinterpret data, provide information that is difficult for humans to comprehend, or overlook details that a human would normally consider. In some cases, AI systems may fail to adapt when faced with situations they were not designed to handle. These issues can lead to poor outcomes, such as incorrect predictions or unfair results, which undermine trust in the technology. This chapter explores the types of AI vulnerabilities, examines real-world examples, and discusses strategies to minimize these risks.
Excerpted from the text; this work has no formal abstract.
Read online
Open full screen ↗Read the paper
The full author-version PDF is free. It opens in your browser's PDF viewer.
BibTeX
@incollection{sawant2026addressing,
title = {Addressing AI Vulnerabilities Through Human-Centered Approaches and Risk Frameworks},
author = {Sawant, Sarvesh and Bhanu, Aasish and Schelble, Beau G. and Madathil, Kapil Chalil},
year = {2026},
booktitle = {Advances in Human-AI Collaboration, John Wiley and Sons},
note = {pp. 325-339},
doi = {10.1002/9781394266401.ch17}
}