WVU professor warns AI forces rethink of college teaching and testing

This is a small, local story that captures a very big global shift: higher education is quietly rewriting its operating system around AI. Farkas’s observation that he can no longer “trust the authenticity of anything that students do unless you watch them” mirrors what many faculty feel but few institutions have formalized. As models like ChatGPT have gone from brittle fact retrievers to competent problem-solvers in math and genetics, they’re colliding head-on with long-standing assumptions about what assignments and exams actually measure.

For the race to AGI, these frictions are less about cheating and more about institutional adaptation speed. Universities are pipelines for both AI talent and social legitimacy; if they respond by banning tools or reverting wholesale to analog assessment, they risk producing graduates misaligned with an AI-native workplace. Conversely, if they move quickly to redesign curricula, assessment and integrity policies around AI-augmented work, they can normalize human–AI collaboration at scale. That cultural normalization is one of the underrated prerequisites for deploying more capable systems down the line.

The piece also highlights how uneven governance still is: policy is being made classroom by classroom, not at the system level. Until universities and accreditation bodies converge on shared norms for acceptable AI use, experimentation will continue to be fragmented, creating both risks and opportunities for tools that can embed guardrails directly into learning workflows.