Navigating Trust and Risk Factors in AI-Assisted Learning Environments: Usage Challenges, Privacy Concerns, and Bias Issues

Research on student adoption of AI tools such as ChatGPT, Claude, and Grammarly is still largely limited to Western adopters, even while these tools race into academic workflows. The very little research conducted on this in South Asia also discourages the dissemination of innovations and such data governance issues as cultural bias within AI outputs, creating a structural limitation that persuades local institutions to choose AI-based communication too. This was a quantitative cross-sectional survey of 705 university students obtained through Google Forms in Pakistan, India, Bangladesh, and Sri Lanka. A UTAUT2 extended framework was used, along with personal innovativeness, usage challenges, privacy concerns, and bias perceptions as context-sensitive constructs. Performance expectancy and personal innovativeness were the most significant positive predictors of behavioral intention. Privacy concerns led to substantially reduced intention. Unexpectedly, perceived barriers and discrimination were positively associated with intention, indicating compensatory uptake. Effort expectancy and hedonic motivation were found to be not significant. In fact, habit in an unexpected way became a hindrance when real use was measured but not the facilitating conditions. Instead of relying on the belief that simply putting in better infrastructure will be enough to encourage users to use it, platform designers and policy-makers should capitalize on users’ capacity for innovation by prioritizing privacy transparency. It is one of the first South Asian cross-national studies extending UTAUT2 to jointly model trust, privacy, and bias perception for AI adoption in a regionally justified way.