Learn faster as your work changes

Listen to the long-form on your commute on Spotify | Listen to short form debate on your commute on Spotify.

The half life of a skill is decreasing (Google Cloud, 2026), and the traditional educational model of 'just-in-case' learning is becoming frail. AI responses closely mirrors the educational level of its input (Anthropic, 2026) My three practical habits that shift AI from answer machine to active learning partner The professionals winning with AI don't focus on the best prompts — they focus on using AI to expand their expertise

This was often me in high school math class. I would learn about the quadratic equation or pythagorean theorem and think — when will I ever use this in life? Perhaps that was you too.

We were taught under an implicit assumption — that acquiring broad knowledge early in life would serve us well decades later. Learn it now, just in case you might need it someday. That model made sense in a more stable world. Today, it feels increasingly fragile.

This shift echoes what economist Steve Levitt described in a Google's Freakonomics podcast earlier this year as “just-in-time learning”: acquiring knowledge when a real problem requires it, rather than storing it speculatively for a future that may not unfold as expected. In a fast-moving professional environment, the practice of just-in-time learning is becoming not just useful, but necessary.

Recent reports highlight how quickly the ground is shifting.

The half-life of professional skills has shortened dramatically — reportedly down to about four years overall, and as little as two years in technology fields (Google Cloud, 2026).

The World Economic Forum’s Future of Jobs Report 2025 suggests that 39% of core skills may need to change by 2030.

In response, the same report ranks curiosity and lifelong learning among the most important capabilities looking ahead — alongside resilience, creativity, and agility.What strikes me is what’s not on that list: prompt engineering or mastery of any specific AI tool. In a world where specific knowledge expires faster than traditional education can refresh it, professional value increasingly comes from our ability to learn and adapt quickly.

Generative AI enters this picture as a powerful enabler of just-in-time learning. Even with a free ChatGPT subscription. But here’s where my thinking gets more nuanced.

When Open AI's GPT 5 model came out, the narrative I noticed was "Its a PhD in your pocket!". Of course, who wouldn't want a PhD level response at their finger tips for $33 CAD per month. But what does a PhD level response actually look like? PhD holders likely know.

Earlier this year, Anthropic's research found a near-perfect correlation (r > 0.92) between the educational level of the input and the education level of the AI's response (Anthropic, 2026). This suggests something important and perhaps uncomfortable: AI doesn’t magically manufacture mastery. It tends to amplify the judgement you already bring to the interaction.

A shallow question usually produces a confident but shallow answer. The tool reflects back the quality of your thinking.

This observation has changed how I use AI. Over the past year, instead of treating it as an answer machine, I’ve started using it as a learning partner — what I call a perspective taker. Rather than outsourcing my thinking, I use it to explore different angles and accelerate my own expertise.

If you want the exact Board of Directors prompt I use, like this post and I'll send it your way.

As a visual learner NotebookLM has been a game-changer that for me. This free tool by Google allows you to upload sources (documents, websites, youtube videos, text, images) to its knowledge sources and you can create videos overviews, slides & infographics podcasts, mind maps and chat with it.

The old just-in-case model is under pressure not because expertise no longer matters, but because knowledge now changes faster than traditional training can keep up.What matters increasingly is learning agility — the ability to quickly identify what you need, absorb it effectively, and apply good judgment.

Generative AI can speed this up. But its real value comes when we treat its output as a first draft and make a thoughtful second pass — refining, testing, and adapting it to our context. Used this way, AI doesn’t replace learning. It helps us become faster, more deliberate learners, prepared for problem's we never anticipated.