|
Getting your Trinity Audio player ready...
|
The University of Rochester just appointed a research computing leader, beginning November 1, to coordinate computing, data, and artificial intelligence. RIT, meanwhile, is enrolling the first cohort in its artificial intelligence bachelor’s program. Together, these investments give Rochester an opportunity to set a useful standard for responsible AI work.
The missing public question is simple: can another person reproduce the result and catch its errors?
The universities should ask student teams using AI in applied research projects to publish a small verification record. Name the input data, identify which claims a human checked, list one discovered model error, and explain what would have happened if it had passed unnoticed. Faculty should grade the quality of these checks alongside the sophistication of the result.
Rochester employers could adopt the same template for AI pilots involving customer service, forecasts, scheduling, or operations. If a result cannot be reconstructed and challenged by someone other than its creator, keep it out of consequential workflows.
New AI programs can advance genuine innovation. Rochester should make transparent error detection part of what its graduates learn to deliver.
Gleb Tsipursky PhD
CEO of Disaster Avoidance Experts and author of “The Psychology of AI Adoption at Work: From Resistance to Results”
The Beacon welcomes comments and letters from readers who adhere to our comment policy including use of their full, real name. See “Leave a Reply” below to comment on this post. Comments of a general nature may be submitted to the Letters page by emailing [email protected].