

Quit Optimizing for Algorithms. Make Something Weird — Robin Sloan
Sep 25, 2025
In this conversation, Robin Sloan, a writer and creative technologist, dives into his unique approach to AI and creativity. He reflects on his early experiences with GPT, noting how AI writing often lacks the depth of human choice. Robin highlights the challenges AI faces with comedy and surprise, while advocating for a return to analog formats like zines for richer connections. He emphasizes the importance of criticism and discovery over algorithmic noise, predicting a future where creators embrace offline mediums alongside digital tools.
AI Snips
Chapters
Books
Transcript
Episode notes
Early Model Training And Burnout
- Robin Sloan trained his own language models around 2016–2018 and built early writing tools using GPT-2-era models.
- He found the technical experiment thrilling but creatively less rewarding, which led to burnout.
The Dearth Of The Author
- AI writing often lacks the accretion of small, individual decisions that give human writing density.
- Max Kraminski calls this problem 'The Dearth of the Author,' highlighting a paucity of decisions in model outputs.
Humor Lives Outside The Distribution
- Comedy and surprise live at distributional edges where AI tends not to go, making human comedians resilient.
- If the punchline sits in the distribution, it ceases to be surprising and ceases to be a real punchline.