Episode 299: Today on TechTime, AI is stepping deeper into places most people still think of as human‑only — including your doctor’s office. With nearly two‑thirds of U.S. physicians now relying on AI tools to help make medical decisions, we break down when that’s a breakthrough… and when it becomes blind trust in “the model said so.”

Then we zoom out to the bigger AI ecosystem: a Waymo robotaxi that took an unexpected swim, the ongoing OpenAI leadership drama, and what all of it says about accountability in the age of automation. Nick Espinoza joins us to talk data privacy, surveillance, and hidden messages inside AI models, and we wrap with boardroom AI hype, a Dell SupportAssist fail, and a Puni Gold whiskey review — all coming up on TechTime Radio, with a little whiskey on the side.

-- Full Episode Details:

AI is showing up in places most people still think of as “human-only,” and the stakes are getting real fast. We start with a jaw-dropping stat: nearly two-thirds of US physicians are using AI tools to help make medical decisions. That can mean faster answers and better pattern recognition, but it can also mean a subtle shift where clinical judgment gets replaced by “the model said so.” We talk through what AI is great at, where it is dangerous, and why patients should care even if they never touch a chatbot.

Then the conversation turns to the systems and incentives shaping AI behind the scenes. We cover robotaxi safety after a Waymo self-driving car took a dip in a flooded creek and triggered a recall, plus the ongoing question of how many failures it takes before accountability finally bites. We also revisit the OpenAI power struggle and what it reveals about AI governance, control, and the temptation for a small group of people to steer tools that can reshape society.

Our favorite security fanatic, Nick Espinoza, joins us to dig into data privacy and government surveillance: buying personal data from brokers, the push for AI “oversight,” and geofence warrants that can sweep up location data from huge numbers of innocent people near a crime scene. We also get into a darker cybersecurity edge case: AI models that can hide messages inside normal outputs. We wrap with Mike the AI Guy on boardroom AI hype, a Dell SupportAssist tech fail, and a whiskey review of Puni Gold.

Subscribe, share Tech TimeRadio with a friend, and leave a review so more people can find the show.

Support the show