AI Can Do the Work, But It Can't Do the Understanding | Better Stack Podcast Ep. 18
Vincent Warmerdam of Marimo joins the Better Stack Podcast to talk Python notebooks, the CoreWeave acquisition, and why LLMs still can't analyze data. He breaks down how Marimo grew 5% a week to become the next evolution of Jupyter, why Python beat every compiled language for data science, and how a gorilla hidden in a dataset exposes the gap between AI doing the work and actually understanding it. Plus mechanical keyboards and RSI, PyPI supply chain attacks, vibe coding and skill atrophy, and the scrubbable IDE he built to make himself care about code again. 🔗 Relevant Links CalmCode YouTube: https://www.youtube.com/channel/UCoi0sH40GlLvNl9TZB40Qeg Marimo YouTube: https://www.youtube.com/@marimo-team Vincent Twitter: https://x.com/fishnets88 ❤️ More about us Radically better observability stack: https://betterstack.com/ Written tutorials: https://betterstack.com/community/ Example projects: https://github.com/BetterStackHQ 📱 Socials Twitter: https://twitter.com/betterstackhq Instagram: https://www.instagram.com/betterstackhq/ TikTok: https://www.tiktok.com/@betterstack LinkedIn: https://www.linkedin.com/company/betterstack 📌 Chapters: 00:00 - Intro & the Keyboard Collection 07:29 - Vincent's Background & Marimo 15:09 - Purple-Pilled on AI 20:48 - The Dynamite Balloon Story 28:26 - Why Python Won Data Science 36:51 - The Road to Marimo 46:30 - CoreWeave & Supply Chain Attacks 50:27 - Vibe Coding & Caring About Code 1:02:19 - Where's the 10x Better Software? 1:12:21 - Vincent's AI Setup 1:15:47 - Turning the Tables on YouTube 1:33:10 - The Gorilla Dataset