A New Vision for Human Work After Automation
Dan Shipper, co-founder and CEO of Every, gives the closing talk from Every's Thesis conference. He argues that as AI handles more expert work, the value of human work moves to the perspective and responsibility people bring to models.
The gist
-
Shipper cites OpenAI's GDPval benchmark: AI beat human experts 12.4% of the time in 2024, 47.6% in 2025, and about 83% in 2026.
-
In one week, Shipper used one Claude subscription to analyze a survey, build a team game, and vibe code a CRM, work he says once took five people.
-
Teammates using the same survey and the same Claude each found different insights, like churn from too much email and readers feeling lonely about loving AI.
-
An Every test found Claude could predict Shipper's Slack replies 93% of the time across 400 guesses, but ranked the right one in its top 10% only 60% of the time.
-
Shipper says default model answers become a baseline everyone shares, so AI should be judged on how well it does your work from your own perspective.
Made by someone on this episode
The Allocation Economy
· Dan Shipper's
— Dan Shipper's 2024 article and framework arguing that knowledge wor...
Also mentioned
◎
Related episodes
Episodes covering similar topics
More from Every
LIVE Vibe Check: Gemini 4 Carbon
Nov 11, 2026
Fable 5.2 Vibe Check
Nov 10, 2026
We Tested Fable 5.2 Early. It’s a BEAST.
Nov 09, 2026
Every Think Week: Monologue HARDWARE?!
Nov 03, 2026