Paper Type
Complete
Abstract
The rapid integration of generative artificial intelligence into organizational and societal contexts has intensified concerns about the reliability and trustworthiness of AI-generated outputs. Among these, the phenomenon commonly termed 'hallucination' remains widely discussed yet inconsistently defined across disciplines. This paper presents a structured survey of AI hallucinations, synthesizing prior research to clarify their evolving definitions, underlying causes, and implications for Information Systems. Complementing this analysis, a bibliometric study of ACM publications from 1995 to 2025 reveals a sharp increase in mitigation-focused research alongside the rise of large language models. We further examine domain-specific implications across healthcare, law, finance, art, and information systems, showing how hallucinations function as both risks and, in some contexts, sources of creative value. Overall, we position AI hallucinations as socio-technical phenomena with direct implications for trust, decision-making, and governance, and provide a foundation for their evaluation, mitigation, and responsible deployment.
Paper Number
1460
Recommended Citation
Thompson, Ramie and Hashemi Tonekaboni, Navid, "Survey of AI Hallucinations and Mitigation" (2026). AMCIS 2026 Proceedings. 12.
https://aisel.aisnet.org/amcis2026/sig_dsa/sig_dsa/12
Survey of AI Hallucinations and Mitigation
The rapid integration of generative artificial intelligence into organizational and societal contexts has intensified concerns about the reliability and trustworthiness of AI-generated outputs. Among these, the phenomenon commonly termed 'hallucination' remains widely discussed yet inconsistently defined across disciplines. This paper presents a structured survey of AI hallucinations, synthesizing prior research to clarify their evolving definitions, underlying causes, and implications for Information Systems. Complementing this analysis, a bibliometric study of ACM publications from 1995 to 2025 reveals a sharp increase in mitigation-focused research alongside the rise of large language models. We further examine domain-specific implications across healthcare, law, finance, art, and information systems, showing how hallucinations function as both risks and, in some contexts, sources of creative value. Overall, we position AI hallucinations as socio-technical phenomena with direct implications for trust, decision-making, and governance, and provide a foundation for their evaluation, mitigation, and responsible deployment.
When commenting on articles, please be friendly, welcoming, respectful and abide by the AIS eLibrary Discussion Thread Code of Conduct posted here.
Comments
SIG DSA