Has AI Conquered Coding? (It’s Not So Simple…) | AI Reality Check
In this episode Cal Newport examines the hype around AI‑driven "agentic coding" by dissecting a recent essay from programmer Lars Faye, who warns that relying on AI to write code can erode developers' critical thinking and skill. Newport highlights real‑world concerns from senior and junior engineers—skill atrophy, fragile code reviews, and new productivity metrics that fuel burnout. Faye proposes a balanced approach: use LLMs for specifications and limited coding, while developers retain hands‑on coding to preserve expertise. The discussion underscores that AI can boost productivity, but only if integrated thoughtfully rather than replacing the craft of programming.
AI Reality Check: Is AI Stealing Entry-Level Jobs?
In this 16‑minute episode of Deep Questions with Cal Newport, the host explores whether AI is displacing entry‑level workers, examining the types of tasks most vulnerable to automation and the skills that remain uniquely human. Newport draws on research and...
AI Reality Check: Can LLMs “Scheme”?
In this episode Cal Newport debunks a sensational Guardian headline about AI chatbots “scheming” by examining the underlying study, which actually tracks Twitter complaints about DIY AI agents built with the open‑source OpenClaw framework. He explains that these agents are...
AI Reality Check: Are LLMs a Dead End?
In this episode Cal Newport examines the claim by AI pioneer Yann LeCun that large language models (LLMs) are a technological dead end and explores the alternative approach of Advanced Machine Intelligence Labs (AMI Labs), which raised over $1 billion to...