The AI Riff (Ep14)

Ep. 14 — Autocomplete, Bubble Talk, and Where the Real AI Value Is Right Now

If “GPT-5 = AGI” was the story you were sold, this episode is your reality check—and your playbook. In 45 minutes, you’ll know what today’s systems actually do well, how to keep them honest, and where the money (and risk) really sits.

Mario DiBenedetto, Rob Katz, Ruth Lee, and Kyle Meyers take the gloves off: Gary Marcus’ latest skepticism, Gartner’s trough-of-disillusionment timing, Sam Altman’s “maybe it’s a bubble” musings, and the week’s safety stumbles from Meta to “spicy modes.”

We zoom from headlines to hands-on—outdated sources, prompt discipline, RAG that actually works, and the platform-vs-app layer knife fight. Plus: our crew’s own builds this week, from real-estate heatmaps and color-tested curb appeal to a live booking app spun up in hours.

Driving question: What should operators expect from AI in 2025—and how do you capture value without getting burned?