The Reliability Gap
Why AI that dazzles in a demo stalls in real work.
Why AI that dazzles in a demo stalls in real work. Shelley and Rob on the distance between a system that can do a hard thing once and a system that has to do an ordinary thing every time. Eight episodes: agents versus assistants, how capability is measured, compounding failure, adoption statistics, press releases versus filings, tracing a famous statistic, missing postmortems, and what happens when the gap closes.
Nothing plays on this site.
- Apple Podcasts Open in Apple Podcasts (opens in a new tab)
- Spotify Open in Spotify (opens in a new tab)
- RSS Feed (opens in a new tab)
Episodes
- 01 What Is It Actually Doing? Runtime 8:43
- 02 The Ruler and the Thing Runtime 8:05
- 03 Nine Times Out of Ten Runtime 6:50
- 04 Eighteen, Forty-One, Seventy-Eight Runtime 7:57
- 05 The Press Release and the Filing Runtime 8:37
- 06 Ninety-Five Percent of What? Runtime 7:53
- 07 The Postmortems That Don't Exist Runtime 9:50
- 08 When the Gap Closes Runtime 9:15