Jeromy, Bogdan and Lee discuss Google integrating AI more deeply into its ad campaign workflow while changing terms so advertisers are liable for AI-generated mistakes, arguing Google gains control without responsibility and predicting potential lawsuits and backlash, especially from professional advertisers. They explore why Google may be pushing this, including internal incentives to grow YouTube advertiser counts, and argue better measurement tools for incremental lift would drive adoption more sustainably. The conversation shifts to an Australian dock workers’ union demanding a 28-hour workweek as automation and AI increase productivity, questioning whether claimed gains are real and where they actually occur. They then cover a Texas Tribune investigation into Texas data centers using permitting tactics for turbines and diesel generators, noting emissions tied to gas plants serving data centers and debating hype versus reality of build-outs. Finally, they discuss Apple reportedly considering acquiring Prism ML to run a compressed 27B-parameter model locally on iPhones, emphasizing privacy and potential market shifts if Apple executes.

00:00 Welcome and Introductions

01:08 Google Ads AI Liability

06:27 Why Google Pushes YouTube

09:20 Measuring CTV Lift Properly

12:09 Australia 28 Hour Workweek

13:14 AI Productivity Reality Check

21:28 Texas Data Centers Emissions

27:59 North Korea Linux Tangent

30:42 Apple Local AI Comeback

37:00 Wrap Up and Subscribe

This podcast is powered by Pinecast.