TBPN Big Tech to Pay for Power, Anthropic Abandons Safety, the Adoption Paradox | Diet TBPN
96 snips
Feb 26, 2026 A fast rundown of contested AI adoption numbers and how surveys may misread real use. A look at who pays for data center power and why communities worry. Discussion of concentrated AI funding, potential mega IPOs, and shrinking public floats. Debate over shifting safety stances at major labs and real-world jailbreaks that expose risks. Quick takes on new AI agents and a surprising crypto investment.
AI Snips
Chapters
Transcript
Episode notes
Measured AI Use Often Doesn't Translate To Perceived Impact
- A rigorous NBER survey shows 70% of firms use AI but 80% report no hiring or productivity impact yet.
- Researchers verified 6,000 leaders across US, UK, Germany, Australia, reducing low-quality self-reporting common in online polls.
Surface Usage Masks Shallow Executive Engagement
- Executives' AI use is widespread but shallow: two-thirds use it regularly yet average only 1.5 hours per week.
- Many employees unknowingly use embedded AI inside SaaS like Toast, so explicit usage metrics undercount adoption.
Invisible AI Means Adoption Is Underreported
- Measuring AI adoption is messy because AI is embedded invisibly in tools and interactions.
- Examples include automated image generation in POS systems and AI-driven call center agents customers don't notice.