Privacy and Trust vs. Utility: Adoption of Commercial vs. Institutional AI assistants Among University Users

Paper Title

Privacy and Trust vs. Utility: Adoption of Commercial vs. Institutional AI assistants Among University Users

Publication Info

• Topic area: AI adoption and user preferences in higher education
• Keywords: Generative AI, institutional AI, commercial AI, trust, privacy, user experience, higher education, task adoption, dual users, AI governance

Background and Problem

• Problem / challenge: Despite universities investing in institutional AI assistants to ensure privacy and compliance, users overwhelmingly prefer commercial AI tools for their daily academic tasks. This raises questions about the utility, trust, and privacy of institutional systems.
• Significance: Understanding the trade-offs between privacy, trust, and utility in AI adoption is critical for designing systems that align with the needs of students, staff, and faculty while maintaining institutional values.
• Motivation and related work: Prior research highlights the advantages of commercial AI in usability and versatility and institutional AI in privacy and compliance. However, comparative studies on how users navigate between these systems for different tasks remain limited.

Solution

• Proposed approach: A survey-based study comparing user perceptions and adoption patterns of institutional and commercial AI assistants at a U.S. university.
• Novelty:
  1. Direct comparison of trust, privacy, and user experience (UX) perceptions between institutional and commercial AI systems.
  2. Analysis of task-specific adoption patterns and dual-use strategies.
  3. Exploration of demographic factors influencing trust and privacy perceptions.
• Procedure and key techniques:
  • Conducted an online survey with 260 participants, including students, staff, and faculty.
  • Measured trust, privacy, transparency, and UX using Likert-scale items.
  • Analyzed adoption patterns across academic and non-academic tasks.
  • Used regression models and qualitative coding to examine relationships between perceptions and usage.

Results

• Concrete findings:
  • Institutional AI was rated higher in trust (M=3.74) and privacy (M=3.00) compared to commercial AI (trust M=2.64, privacy M=2.15).
  • Commercial AI dominated in usability and task adoption, particularly for writing, coding, and research assistance.
  • Trust in institutional AI was linked to institutional governance, while trust in commercial AI was tied to experiential quality (e.g., satisfaction, response quality).
• Advantage over baselines:
  • Institutional AI was preferred for privacy-sensitive tasks (e.g., administrative work), while commercial AI was favored for productivity-oriented tasks due to speed and versatility.
• Experiments / evaluation:
  • Surveyed 260 university users, with 43% using only commercial AI, 17% using only institutional AI, and 26% using both.
  • Statistical tests (e.g., Mann–Whitney U, Wilcoxon signed-rank) and regression analyses identified significant differences in trust, privacy, and UX perceptions.
• Limitations and future work:
  • Limited faculty participation and single-institution context may affect generalizability.
  • Self-reported data may not fully capture actual usage behaviors.
  • Future research should include cross-institutional studies, longitudinal analyses, and experimental interventions to enhance transparency and usability.

Summary

This study reveals a trade-off between trust/privacy and utility in the adoption of institutional and commercial AI assistants in higher education. Institutional AI is trusted for its compliance and governance but lags in usability, while commercial AI dominates in productivity tasks due to superior user experience. Dual users strategically allocate tasks between systems based on context. The findings highlight the need for institutional AI to enhance visibility of privacy protections and for commercial AI to address transparency gaps. Universities should focus on complementing commercial systems rather than competing directly, leveraging their strengths in governance and accountability.