Key Takeaways

  • Learning splits into three core types: knowledge (lectures), skills (exercises), and wisdom (discussion/interaction).
  • AI brings a novel way to gain "wisdom" through conversational learning, a capability traditional methods lacked.
  • Despite AI's promise, Matt Pocock found that "AI experimental" learning approaches actually deter learners.
  • The most effective method, especially for technical subjects like TypeScript, remains a "problem-first" approach where learners tackle challenges before receiving explicit knowledge.

The Three Pillars of Learning, and Where AI Fits In

Matt Pocock, known for his work in TypeScript, has a clear hierarchy for how humans acquire expertise. He breaks learning into three distinct categories: knowledge, skills, and wisdom. “You can teach them knowledge, and you do that through lectures, right?” he explained on Latent Space. This is the foundational layer, the "what." Next come skills, the "how to do it," which Pocock says you build "with an interactive exercise." But the real prize, the "why," is wisdom—a deeper intuition gained through experience and discussion.

Here's where AI makes a surprising entrance. Pocock notes that for wisdom, “working with AI is something you can can sort of get to wisdom a little bit by talking to AI. You've never really been able to have that with any kind of learning mechanism before, you know?” This isn't about AI delivering knowledge or showing you a step-by-step. It's about a conversational partner that can challenge assumptions, offer different perspectives, and simulate scenarios, helping you forge that hard-won understanding.

Why 'AI Experimental' Approaches Fall Flat

Given AI's potential for wisdom, you might expect Pocock to be all-in on AI-driven learning platforms. But his experience tells a different story. He's observed a clear paradox: “what I've noticed is the more I lean into the kind of AI experimental stuff, the more it actually turns people off my materials.” It seems that while the concept of AI as a wisdom-generating tool is exciting, its practical application in core learning still struggles to resonate with learners.

The allure of cutting-edge tech doesn't always translate to effective pedagogy. Instead of embracing every new AI learning gadget, Pocock found that people still prefer the proven, if less glamorous, path. They want clear structure, challenging problems, and a sense of earned accomplishment that shiny AI tools often dilute. For founders and builders, this is a signal: innovation in learning doesn't always mean layering on more AI; sometimes it means doubling down on what truly works for the human brain.

The Problem-First Advantage

So, what does work? For Pocock, it’s a rigorous, problem-first methodology. Forget lectures as the starting point. His approach flips the script entirely. “I'm very careful with my stuff to really force people to do the work, essentially,” he said. He designs his courses, like his popular TypeScript materials, to throw learners directly into a challenge. You encounter a problem, you struggle with it, and then you get the knowledge needed to solve it or understand the solution.

This isn't about trial and error for its own sake. It’s a deliberate design choice to build stronger mental models and deeper retention. By grappling with a problem first, your brain becomes primed for the information. The knowledge then serves as a tool to overcome a tangible obstacle, rather than just abstract facts to memorize. “Even in my TypeScript stuff, I would structure it so that people would be thrust into a problem and have to solve it on your own. I would give you the knowledge afterwards,” Pocock detailed. This method bypasses the passive consumption of lectures, forcing active engagement from the outset.

What to Do With This

Stop waiting for AI to magically upskill your team or yourself. This week, pick a critical skill your team needs (e.g., advanced debugging, specific architectural patterns). Instead of sending them a YouTube lecture or asking them to query an AI, frame it as a live, urgent problem they must solve. Give them minimal context, force them to experiment and fail, and only then provide the core knowledge. You'll build deeper skills and genuine wisdom, not just memorized facts.