A learning philosophy for the AI era

Optimistic about AI. Pragmatic in approach.

The future won't be built by the hyped or the terrified. It will be built by people who believe in what AI makes possible — and do the unglamorous work of learning it honestly.

Between the doomers and the zoomers there is a quieter position: that AI is a genuinely transformative — and genuinely normal — technology, whose benefits go to the people who engage with it deliberately. Because the hard part of AI was never the technology. It's people — and people don't install. They learn.

The six stages

Learning AI is a journey, not a download.

Every meaningful skill follows an arc — curiosity, conviction, struggle, absorption, insight, fluency. This is that arc, mapped for the AI era and grounded in learning science — with one practice to try at every stage.

Stage one

Spark of Curiosity

It starts with a jaw-drop moment — the first time an AI does something you didn't think was possible. Curiosity is an information gap: the space between what you know and what you suddenly want to know. Don't let the spark pass. Follow it.

In practice: reproduce the jaw-drop within a day — take whatever amazed you and try it on your own real work while the spark is hot.

Grounded in Loewenstein's information-gap theory of curiosity (1994) and Hidi & Renninger's model of triggered interest (2006). Sources →

Stage two

Ethical Crossroads

Before you go deep, decide who you want to be with this technology. What will you use it for? Whose work does it touch? Where are your lines? Ethics isn't a module you bolt on at the end — it's the compass you pick up at the start.

In practice: write your three lines — what you'll always use AI for, what you'll never use it for, and what you'll always disclose. Then tell one person.

UNESCO's 2024 AI Competency Framework puts a human-centred mindset and ethics first, before technique — and MIT's AI literacy curricula interleave ethics with every lesson. Sources →

Stage three

Challenge Labyrinth

Then it gets hard. Prompts fail, outputs disappoint, the hype doesn't match your screen. This is the stage where most people quit — and it's exactly where the learning happens. Difficulty that slows you down today is what makes the skill stick tomorrow.

In practice: keep a dead-end log — one line per failure: what you tried, what happened. Ten entries in, the pattern shows itself.

Bjork calls these "desirable difficulties" (1994); Kapur's "productive failure" research shows struggling before instruction beats instruction alone. Sources →

Stage four

The Rabbit Hole

Somewhere in the struggle, the switch flips. You lose track of time. You're generating your own questions now, chasing threads nobody assigned you. Situational interest has become personal interest — the rabbit hole is where a curiosity becomes yours.

In practice: book the deep block — ninety protected minutes on one problem you actually care about. Reps, not tabs.

This is Csikszentmihalyi's flow state — total absorption when challenge meets skill — and the shift to "well-developed individual interest" in Hidi & Renninger's model. Sources →

Stage five

Breakthrough

The aha moment is real — impasse, then sudden restructuring, then certainty. In AI work, the breakthrough is usually the same discovery: you've internalized the jagged frontier. You know, without checking, where AI is brilliant and where it will quietly fail you.

In practice: write the frontier memo — one page on where AI is brilliant for your work and where it quietly fails. Date it; it will change.

Kounios & Beeman mapped the neuroscience of insight (2014); Harvard's "jagged frontier" study documents the AI version — consultants thrive inside the frontier and stumble just past it. Sources →

Stage six

Mastery

Mastery isn't a certificate — it's fluency you no longer have to think about, kept sharp by structured practice and feedback. And in the oldest arc we know, the journey ends with a return: the master's last responsibility is to hand the elixir to someone at stage one.

In practice: teach one thing within a week — show one person one real workflow. Teaching is how fluency sticks, and it's how the cycle restarts.

Ericsson's deliberate-practice research (1993) and the Dreyfus skill-acquisition model define the ladder; Campbell's hero's journey supplies the return. Sources →

Guiding someone else through these stages? That's the other seat — the coaching cycle.

“Optimism supplies the reason to start. Pragmatism supplies the way through.”
— from the Optimistically Pragmatic manifesto

Read the manifesto

Standing on shoulders

Built on ideas that already work.

Optimistically Pragmatic didn't appear from nowhere. It extends a lineage of product and design thinking — empowered teams, story mapping, goal-directed design, usability — into the AI era.

Product

Cagan & Patton

Empowered teams solving problems that matter, and story maps that build shared understanding before building software.

Design

Cooper & Nielsen

Design for real people with real goals — and never mistake what's technically impressive for what's actually usable.

Coaching

Wooden & the court

Because AI adoption is a people game, the playbook comes from the great coaches: define success as growth, feed people information instead of judgment, and turn your masters into coaches.

Read the coaching philosophy →  ·  Explore the influences & research →