It's a Skill, Not a Pill
I was wrapping up a keynote at Wharton earlier this week. I'd just walked through a Wharton / Harvard study of hundreds of BCG consultants — AI users outperformed non-users by 40% on quality. (Not 4%. Forty.) Then I asked the room a question I've now asked in hundreds of rooms:
If there were a pill that boosted your IQ by 30 points with no side effects, would you take it?
Of course you would. It's a no-brainer.
Except — if it's such a no-brainer, why aren't more people taking it? Section's 2026 AI Proficiency Report found that less than 3% of the workforce are AI practitioners or experts — people who actually put AI to use in their workflows and see significant productivity gains. Two years ago, that number was 9%. A year ago, 7%. It's going the wrong direction.
A 40% quality boost, available to anyone. Taken up by fewer than 3%.
The pill is sitting right there. Almost nobody's swallowing it.
So I said what I often say to close: I want you to take the 30 IQ point pill.
And then I heard myself add: "And it's not really a pill, to be clear. It is a skill."
I paused. That's a really good phrase. I wish I'd thought of that before. (I didn't plan it. It just came out. Which is usually how the best things happen.)
The wrong metaphor
My last few posts have been about the other end of this problem — what happens when you've unlocked AI capability and it starts consuming everything else. (If you read "Are You More of What You Value?" or "The Rising Opportunity Cost of Being Human," you know what I mean.) Those posts were for people in danger of using AI too much — where productivity swallows presence, and you look up one day and realize you optimized the wrong variable.
But today I want to talk about the starting line. Because the uncomfortable truth is that most people aren't in danger of using AI too much. They're in danger of never starting.
And the pill metaphor is part of the problem.
Pills work instantly. You swallow them, they work. Skills don't work that way. Skills require a progression — from fear, to familiar, to fluent, to fun. They require reps. They require doing it badly before you do it well. When people treat AI like a pill they were supposed to have already taken, of course they feel shame when they haven't. Of course they aren't enthusiastically elbowing their colleagues to share what they're doing. The pill framing sets up a pass/fail test where most people feel like they've already failed.
The skill framing gives them a path.
The permission to disengage
I was on a panel last week before a gathering of private equity investors. A fellow panelist said something well-meaning: "You can't become an AI expert in an afternoon…"
You could see it happen in real time. A kind of permission to disengage rippled through the crowd. I'm a hospitality expert. I'm a transportation expert. I'm an oil and gas expert. I don't need to become an AI expert.
I jumped in. "I think a lot of you just heard that and concluded that you don't need to become an AI expert. And nothing could be farther from the truth."
Then I asked the room a question.
How many of you would call yourself an internet expert?
No hands.
How many of you would say you're an email expert? A spreadsheet expert?
Nervous laughter.
"You are," I told them. "You just don't think about it that way — because it's a skill you built so gradually you never noticed you were building it."
That's the point. Nobody woke up one morning fluent in email. You drafted, sent, replied, learned what worked, figured out when to CC and when not to. (Remember when you used to print emails? I do. I'm not proud of it.) You built invisible competence over years of daily practice.
AI is the fourth foundational utility of the modern knowledge worker, alongside electricity, devices, and internet. Saying "I don't need to become an AI expert" is the same as saying "I don't need to become an internet expert." It reveals a belief that AI isn't indispensable.
It is.
What skill development actually looks like
My youngest daughter was learning to ride a bike. Every time she fell, she'd get more tense. More afraid. More reluctant to try again.
So I reframed the goal. Your job today is to fall. That's it. See how many times you can fall.
She started laughing. The tension broke. And she got on the bike in a completely different spirit — and started succeeding far sooner than she would have otherwise.
Albert Bandura, the Stanford psychologist, studied this mechanism his entire career. His concept of guided mastery showed that small, sequential wins don't just reduce fear of a specific thing — they generalize. Subjects who tackled one feared challenge with graduated support didn't just lose that one fear. They stopped seeing new challenges as threats entirely. Self-efficacy — your belief in your ability to handle difficulty — transfers.
The implication for AI is direct: you don't need to master AI. You need a series of small wins that build your belief that you can get better. Each one expands what feels possible. And the expansion is the whole game — because once people start believing they can figure this out, they start finding use cases I've never thought of (those examples are always my favorites!). That's when I know the skill has taken hold.
Daily practice is the inflection point
Bryce Challamel (former Head of AI at Moderna — one of the most sophisticated AI-deploying organizations in the world — now at OpenAI) describes four buckets of AI users he has studied:
Haven't really engaged with the technology
Using it like an improved search engine (one-shot prompts, one-shot answers)
Daily users making steady progress, optimistic about learning
Advanced users flying with it — agentic workflows, real leverage
His observation: the critical inflection isn't between buckets 3 and 4. It's between 2 and 3. Daily practice is what creates momentum.
James Clear cites research showing that 1% daily improvement compounds to 37x over a year. (1% daily decline? You lose 97% of your capability.) That's the math behind what Bryce is observing — the gap between bucket 2 and bucket 3 isn't talent or technical aptitude. It's daily reps.
Brice put it memorably: using AI is like showering. If you don't do it every day, it's a problem — and at some point, there are diminishing returns. (I laughed when he said that. But I also haven't stopped thinking about it.)
Brad Anderson, president of Qualtrics, said something similar on my podcast: "If a company wants to lead in the AI age, every employee has to use AI every single day. Every employee. Every function. Every day."
That's not a high bar. It's a different bar. Not how much but how consistently.
What to do (starting today)
The way to build this skill is not to study it. It's to do it.
1. Pick one task you do every week and try it with AI today. Not your most critical, career-defining project. Something low-stakes. A weekly summary. A meeting prep. A first draft of something routine. (The lower the stakes, the more you'll actually experiment. That's the bike-riding lesson.)
2. Make it daily — even if it's 10 minutes. The Moderna/OpenAI data is clear: the leap that matters is from "occasional user" to "daily user." Put it on your calendar. Literally. Block 10 minutes. The quality of what you do in those 10 minutes matters less than the fact that you showed up.
3. Redefine success as reps, not results. Your first attempts will be mediocre. Good. That's not failure — that's the process. My daughter's job was to fall. Your job is to prompt badly, get weird outputs, learn something, and try again tomorrow. Bandura's research is clear: the wins compound and the confidence transfers.
4. If you want a structured path, enroll in my personal AI bootcamp. 15 days. 10 minutes each. Designed to expand the surface area of what feels possible. Each drill isn't the point — the point is that after 15 days of small wins, you'll start imagining use cases even I haven't thought of.
The 30 IQ points were never in a pill.
They were always in the reps.
Related: You're Not Tired of AI. You're Scared of It.
Related: Are You More of What You Value?
Related: The Rising Opportunity Cost of Being Human
Join over 31,147 creators & leaders who read Methods of the Masters each week.
A 40% quality boost, available to anyone. Taken up by fewer than 3% of the workforce (and that number is going down, not up). At Wharton this week, I accidentally said the thing that explains why.