In the latest episode of The AI Values Podcast, host Edosa Odaro sits down with Jon Cooke, a veteran with 20 years of experience inside AI systems, to tackle one of the most persistent problems in artificial intelligence: hallucinations.
Cooke, who has seen the evolution of AI from the inside, warns that many executives believe hallucinations are already solved—a notion he dismisses as a sales pitch. "I’ve been in rooms with executives who think hallucinations are already a solved problem," Cooke says. "Yes, there are vendors telling them exactly that."
The conversation dives deep into the technical and practical challenges of eliminating AI hallucinations, offering insights that go beyond surface-level fixes. Cooke argues that true solutions require understanding the underlying systems, not just applying patches.
For those interested in the full discussion, the complete episode is available at bit.ly/4t33HpT. The AI Values Podcast aims to bring clarity to complex AI topics, cutting through hype to focus on real-world impacts.
Tune in to hear whether Jon Cooke has actually solved AI hallucinations—or if the industry still has a long way to go.