
AI and the Future of Programming

Merchants of Complexity 🏯 — with DHH
Topic 2 of 3
Published Sep 5, 2024 · Clip 34:50–48:54 (14:04)
David describes how he uses LLMs as a non-driving pair programmer, loving quick answers but warning that reliance erodes skill muscles—illustrated by his Amakub Ubuntu remix built in Bash, where LLM answers took longer to sink into muscle memory. He notes LLMs are 'the good Google' since search has degraded. He reflects on genuinely enjoying programming like playing piano rather than consuming Spotify, wonders if programming may become like horseback riding, and argues we may be at a hype-cycle peak, comparing AI to self-driving car proclamations, VR, the Segway, and the jet-flight ceiling. He concludes progress can't be paused and productivity gains are net positive, advising engineers to take off the training wheels periodically.