Issue 148  July 3rd 2026

AI is improving your work. It may be weakening you.

New research puts hard data behind the deskilling risk. Plus: a 10-minute practice to protect the skills that make you valuable.

THE ARGUMENT

You finished the deck in 20 minutes instead of two hours. It was good. But could you have built it without the tool? Be honest.

That question just stopped being rhetorical. The evidence arrived this month - and it's not subtle.

Nature reviewed the first wave of empirical studies on AI-driven deskilling. The early results: reliance on AI tools can measurably degrade the abilities of professionals, including physicians and software engineers. Not hypothetically. Measurably. And if it's showing up in medicine and engineering, it's showing up in product, design, and strategy too.

A WSJ Intelligence study named both sides of the ledger at once - a "Human Premium" for skills judged hard to replicate, alongside an explicit warning about cognitive atrophy as corporate automation accelerates.

And PwC's analysis of a billion job postings showed the market is already pricing the gap: roles that amplify human judgment are growing at twice the rate, with 42% faster wage growth, than roles that merely cut cost.

This isn't a prediction anymore. Deskilling is documented. The market is responding. The question is whether you are.

The six undelegatable skills - Curiosity, Critical Judgment, Empathy, Advocacy, Collaboration, Experimentation - aren't a nice framework. They're the specific capabilities the data says are hardening in value while everything around them gets cheaper.

AI can improve the work without improving the worker. That's the trap I call synthetic competence: polished output running ahead of human judgment. The work looks sharp. The person underneath may be getting fewer reps in the very skills that make them valuable.

THE TELL

Here's how to catch it in your own work this week. (And if you recognize these on your team, forward this to a colleague so you have a shared vocabulary for it.)

• 🚩 You can't explain your recommendation without reopening the AI output.
• 🚩 Your first instinct on a new problem is to prompt, not to think.
• 🚩 The time you "saved" with AI went to more tasks, not deeper judgment.
• 🚩 Your team produces more deliverables but defends fewer decisions.

If the tool disappeared tomorrow, would the skill remain? That's the undelegatable test.

YOUR PRACTICE THIS FORTNIGHT

🛠 The Human-First Rep (no AI - do this first)

Pick one professional skill you've quietly handed to AI in the past year — writing, analysis, synthesis, research. This fortnight, do one instance of that task manually. Start to finish, no AI assist.

Notice what felt harder. That difficulty is diagnostic: it's the muscle that's been losing reps. Make this a recurring ritual for the 2–3 capabilities you most need to keep sharp.

Then ask yourself: If I had to defend this work in a room without the artifact, could I?

🤖 The Sparring Prompt (now bring AI in - as a challenger, not a crutch)

After you've done the manual version, open your AI tool and paste this:

"I just completed [task] without AI assistance. Here's my output: [paste it]. Now act as a critical reviewer: What did I miss? Where is my reasoning weakest? What assumption am I not questioning? Don't improve my work - expose where my thinking is thin."

Compare what the AI flags to what you noticed yourself. The gap between the two is your atrophy risk — the judgment you're close to outsourcing.

This is the PAIR loop in practice: Prime (you did the thinking first), Assess (AI challenges your work), Interpret (you decide what to act on), Retain (the skill stays yours).

Reply and tell me: which skill did you pick for your Human-First Rep? I read every response.

THE PROOF (what I read that backs this up)

📄Is AI Ruining Our Skills? Early Results Are In - Nature, June 18 The first empirical studies on AI deskilling. Physicians and engineers measurably degraded. This is the piece that moved the conversation from "could happen" to "is happening."

📄 WSJ Intelligence: The "Human Premium" - Yahoo Finance / WSJ, June 23 Names cognitive atrophy as a corporate risk in the same breath as the human-skills premium. Both sides of the ledger, in one study.

📄 AI Is Splitting the Job Market - Straits Times / PwC, June 16 Across 1B+ job postings, roles amplifying human judgment grow 2x faster. The market is pricing what you practice.