Deeply Human Leading with Lee Ann Del Carpio & Gretchen Terry-Leonard
When AI starts learning mostly from itself, where does anything genuinely new come from? Anthropic reported that more than 80 percent of Claude's code is now written by Claude, and a premier model vanished from the team's plans overnight. Lee Ann and Gretchen use the humble banana to explain AI monoculture and model collapse: almost every banana we eat is a Cavendish, a near genetic clone that one disease could wipe out. They get into the reflex to overtrust machines and undertrust humans, how quality quietly erodes from good to "it'll do", and what the AI itself said when asked how it avoids becoming a clone of a clone. The throughline: diversity is insurance you have to design in, across models, money, and the counsel you trust. CHAPTERS: 00:00 Two things that knocked us offside 02:29 Why Gretchen's mind went to bananas 02:43 The Cavendish problem: one banana, one disease 05:31 You can't run a business on one banana 06:44 Blind spots and the autocorrect bias 08:21 Overtrust the machine, undertrust the human 09:22 Running out of naive data: getting high on your own supply 10:12 Where the AI constitution fits in 11:56 Asking the AI to confront itself 14:00 Variance has to come from outside the loop 15:47 Diversity is insurance, never a guarantee ABOUT THE HOSTS: Lee Ann Del Carpio is CEO and Founder of Deeply Human and Founder of the Center for Deeply Human Leadership. Gretchen Terry-Leonard is Co-Founder and Chief Strategy and Impact Officer of CDHL and author of Your Second Prime. WEBSITE: deeplyhumanleading.com [http://deeplyhumanleading.com] NEWSLETTER (The Deep Signal): https://deeply-human.kit.com/b1bfa01738 [https://deeply-human.kit.com/b1bfa01738]
13 episodes
Comments
0Be the first to comment
Sign up now and become a member of the Deeply Human Leading with Lee Ann Del Carpio & Gretchen Terry-Leonard community!