Opinion: UMass’ AI Guides Should Test What Students Can Verify
Photo: Shutterstock
The UMass Amherst AI Guide Network is recruiting faculty, teaching staff, and librarians through October 23. Its strongest contribution would be a common standard for showing what students can independently judge after using AI.
That standard is overdue. Western Governors University’s September 30 survey found that 60% of 3,128 U.S. hiring professionals say AI makes it harder to assess candidates’ real skills. The share still figuring out how to assess AI skills doubled from 16% last year to 32% this year.
UMass should let instructors devise assessments that require students to challenge an AI answer. A public-policy student could identify a missing stakeholder in a draft memo. A biology student could catch a fabricated citation and rebuild an explanation from the underlying data. A business student could document where an AI forecast breaks down and defend a corrected recommendation.
Collect a small portfolio showing the original output, the student’s tests, the corrections, and the final judgment. Have instructors score the reasoning, then ask employers which demonstrations matter for entry-level work.
The Guides’ disciplinary expertise makes them well placed to create these tests. Amherst students deserve credentials that show employers something more informative than how fluently they can prompt.
Gleb Tsipursky, is a behavioral scientist, CEO of Disaster Avoidance Experts, and author of The Psychology of AI Adoption at Work: From Resistance to Results (Georgetown University Press, 2026)