Weekly Newsletter

Practical AI Strategies

This week on the blog... Accessibility and Expertise

Plus: Easter course discount ends soon

Hi everyone,

Two posts this week, both circling the same question: what actually reaches people when knowledge travels through a medium?

The AI industry's implicit pitch is that the medium has finally become transparent. You put knowledge in one end and get it out the other, perfectly preserved, infinitely scalable. Both of these posts push back on that idea, from different angles.

IYKYK Part 3: Who Gets to Know?

This is the third post in the IYKYK series, which started by arguing that GenAI has a "discoverability problem": the interface gives users no signals about what's possible, so most people get stuck in 2023-mode. In this post, I argue that the discoverability problem is also an access problem, and one that's been chosen deliberately by the companies building these platforms.

Every major AI platform fragments its capabilities across pricing tiers and different apps. Free users hit rate limits and think "AI isn't very useful." Schools with bigger budgets get access to more capable tools, build stronger mental models, and develop more sophisticated workflows. Schools without those resources hit walls earlier and file AI under "not that useful." Meanwhile, the most capable AI available right now isn't the chatbot at all; it's command line tools like Claude Code that are almost completely out of reach for schools.

I also make a practical suggestion: give a small group of enthusiastic educators a sandboxed device loaded with the best available tools. It costs less than sending three staff to a conference, and it gives you a permanent capability within the school.

Expert Signals: Fidelity

For eighteen months I took weekly jazz guitar lessons over Skype. My teacher had an instinct for my energy I never had to explain. I also learned from Justin Guitar's structured video courses, which have taught millions to play. And I watched John Mayer walk through his songs on Instagram during lockdown. Three sources, three very different channels, and the fidelity of what reached me had nothing to do with who was the most accomplished guitarist.

This post uses signal processing as a lens for knowledge transfer: what survives compression, what gets stripped out first, and why AI discards exactly the components of expertise that make it worth receiving. A lossy signal from an expert is still enormously more valuable than a lossless signal that contains no tacit knowledge at all.

Cheers,

Leon

PS: All courses and digital downloads at Practical AI Strategies are 25% off over the break. Use the code EASTER-2026 at checkout. Ends April 26th.

Stay informed.​
The Practical AI Library curates fresh, practical articles, research, and advice on AI and education—updated every week. Check it out
Check it out → Practical AI Library​

​​​