Silo Busting 62 Human Testing and Generative AI with Tariq King and Ira Livshits

The era of generative AI needs humans… not just the data sets they create. Now and going forward, companies will need to depend on perceptive, intelligent, creative humans who can evaluate and, when necessary, stand up to the black box output AI throws at them. What kind of traits do such people embody? Where might we find them? What might they teach the rest of us? That’s what this episode of *Silo Busting* is all about. Tariq King, our VP and Head of Product-Service Systems, and Ira Livshits, Senior Data Scientist at EPAM, meet the test-based questions of Producer Ken Gordon with knowledge and aplomb. King, who spent much of his career hiring testers, says he looks for “the way that someone actually approaches solving problems and thinking perhaps in a different way or from a different perspective.” Livshits notes that with testing in generative AI, “sometimes there is no right or wrong” and that a tester will “have to act or decide based on his or her understanding of the field and maybe intuition.” In short, she says, testing has become much less of a black-and-white job. Together King and Livshits create a model conversation. They talk about how explainability fits in here, why musicians and liberal arts people make for good testers, the importance of being able to relate findings to a given audience (“[if] I have a PhD focused on machine learning, probably I can understand certain things in a different way than someone who may not have that degree,” says King), and how generative AI is bringing the role of testing closer to the fore. Go on, give it a test-listen. Host: Alison Kotin Engineer: Kyp Pilalas Producer: Ken Gordon