The investor class is finally asking the question knowledge workers have been avoiding: what happens when thinking becomes optional?
The Summary
- Tom Slater of Scottish Mortgage Investment Trust argues AI's real threat isn't job displacement but cognitive atrophy — a world that appears productive while quietly eroding human judgment and expertise
- The risk: we optimize for output while accidentally destroying the learning loops that build real capability
- Slater's paper reframes the AI debate from "will I have work" to "will I still know how to work"
The Signal
Tom Slater manages billions at one of the world's most forward-looking investment trusts. When he publishes a paper called "AI Isn't Coming for Your Job. It's Coming for Your Mind," the market listens. His argument: we're so fixated on whether AI will replace jobs that we're missing the more insidious threat.
The real risk is a productivity trap where systems look efficient on dashboards while the humans operating them gradually lose the skills that made them valuable in the first place.
"A world that looks more productive while quietly losing the judgement, learning and expertise that make progress possible."
Think about junior lawyers who never draft their own briefs because AI does it faster. Or analysts who stopped building financial models from scratch because the agent already ran the numbers. The work still gets done. Probably faster. But the learning loop that turns juniors into seniors, that builds intuition and judgment, that loop quietly breaks.
Slater's framing hits different because he's not a tech skeptic or a luddite academic. He's a growth investor who's made fortunes betting on automation and AI companies. This isn't fear. It's pattern recognition.
Key risks Slater identifies:
- Skills atrophy: When AI handles routine tasks, humans never develop the foundational competence needed for complex judgment
- Invisible dependency: Organizations become reliant on systems they don't fully understand
- Innovation stall: Breakthroughs require deep expertise, which requires years of deliberate practice, which AI short-circuits
The workforce implications are stark. We've spent two years debating which jobs disappear. Slater is asking what happens to the jobs that remain when the people doing them never learned to think without the assist. A senior engineer who started coding with Copilot might ship features faster but struggle to debug novel problems. A doctor who diagnosed with AI from day one might miss the subtle patterns that models haven't been trained on.
The Implication
This reframes the entire AI skills conversation. It's not about learning to prompt engineer. It's about deliberately designing friction back into learning processes so people still build real capability. Companies optimizing purely for output are accidentally training a generation of workers who can operate systems but can't build them, improve them, or work around them when they fail.
For investors, Slater's pointing at a less obvious opportunity: the tools and training systems that help humans maintain expertise in an AI-mediated world won't be productivity plays. They'll be insurance against institutional brain drain. Watch for companies building "deliberate practice" into AI workflows, not just speed.