What Happened
Casey Muratori, the lead programmer at Molly Rocket and creator of the Computer Enhance programming course, launched a new discussion series called "Wading Through AI." The first episode, "Should You Be a Carpenter?", went live on March 6, 2026.
Muratori co-hosts the series with Demetri Spanos, an independent researcher in computational mathematics and scientific computing who runs a consultancy focused on commercial ML, AI, optimization, and robotics. The two are longtime friends, and the series grew out of Spanos being constantly asked by people in his life to make sense of AI developments.
The central question of Episode 1: is entering the knowledge economy today too risky, given how fast AI is moving? The title isn't really about carpentry. It's about whether physical, hands-on trades might be a safer career bet than the white-collar knowledge work that AI tools are increasingly capable of doing.
The series is published through Muratori's Computer Enhance Substack. Notably, Muratori states explicitly that Molly Rocket does not use generative AI, and all artwork for the series was hand-drawn by Anna Rettberg.
Why It Matters
This isn't another "AI will replace all jobs" hot take or a "don't worry, AI is just a tool" reassurance piece. What makes this series worth paying attention to is who's involved.
Spanos has been working in AI research for decades, long before the current LLM wave. Muratori is a deeply technical programmer known for rigorous, no-nonsense analysis (his "Clean Code" critique and Handmade Hero series have a strong following among developers). Together, they're positioned to have a more grounded conversation than most.
The "should I become a carpenter" question is one that a lot of people are actually asking right now, especially junior developers, students, and early-career knowledge workers. The tools we review on this site - Cursor, Claude Code, ChatGPT - are genuinely changing what one person can produce in a day. That productivity gain is real, but the career implications are complicated and mostly unresolved.
For anyone using AI productivity tools daily, this is the uncomfortable background question: am I building skills that compound, or am I training alongside my replacement?
Our Take
We've needed more voices like this in the AI conversation. The current discourse is dominated by two camps: AI boosters who treat every model release like a second coming, and doomers who predict civilizational collapse. Neither is useful for someone trying to make practical career decisions in 2026.
Muratori and Spanos are explicitly trying to avoid both traps. Muratori describes Spanos as someone who "never tries to sell a specifically positive or specifically negative view of AI" and instead provides technical context so people can draw their own conclusions. That's rare.
The "carpenter question" is a smart framing because it forces you to think about what AI actually can and can't automate. Physical trades require embodied skills, spatial reasoning, and real-world problem solving that current AI doesn't touch. Knowledge work - writing, coding, analysis - sits squarely in the zone where LLMs are most capable.
That doesn't mean everyone should go buy a toolbelt. But it does mean knowledge workers should be honest about which parts of their work are most vulnerable to automation, and which parts require judgment, taste, and context that current tools still can't replicate. If you're using AI tools effectively, you already know the answer is nuanced. This series looks like it'll treat that nuance seriously.
Worth subscribing to if you want a technically informed perspective that isn't trying to sell you anything.