Is the technology acceptance model still useful for generative AI?
The technology acceptance model is one of the fastest-growing keywords across the tracked journals, and a quarter of recent TAM and UTAUT papers now study AI. But the model was built for workplace software decades ago. Do perceived usefulness and ease of use predict who actually uses generative AI, and what do they miss?
What prompted this
"Technology acceptance model" grew across all journals (19 papers in July–September 2026 against about 10 expected) and in finance (8 against 2); see the trends page. Since October 2025, 109 papers carry a TAM or UTAUT keyword (67 in management journals, 24 in finance); 26 of them also carry a generative-AI, ChatGPT, language-model or AI keyword.
Questions
- Do intentions measured with TAM surveys predict actual use, as logged by the tools, weeks later?
- Do things the model leaves out, such as trust in accuracy, fear of replacement or the tool doing the whole task, explain more of who uses it?
- Do results replicate across countries and occupations, or reflect the student samples many studies use?
Possible approach
Pair a standard TAM survey with usage logs from an AI tool deployed in a firm or university, and compare its predictive power with and without the added factors.
Next step
Survey the users of one AI tool with a standard TAM questionnaire and match their answers to their usage over the following month.