In 1806, a Lancashire handloom weaver earned 20 shillings a week. By 1830: six. The power loom didn't replace him - it repriced him.
AI is repricing knowledge work on the same pattern - but with a twist. The work being automated is the same work that trains the next generation of experts. Automate the rungs, and the ladder disappears. Call it the rung gap: the growing void between entry-level work that machines now handle and senior judgment that still requires a decade of experience to develop.
What does the repricing actually look like?
When a machine can do what you do, your pricing power collapses to the gap between its output and yours. For a junior lawyer drafting research memos, that gap is narrow and narrowing - GPT-5 produces competent legal research in minutes. The junior's value was never in the quality. Seniors always rewrote it. It was in the labour. That labour is now automated.
For a senior partner who knows when to settle, how to read a judge, which argument survives appeal - the gap is wide. These are judgment calls built from decades of pattern recognition. The partner's leverage increases when intelligence is cheap, because she can deploy more of it, faster.
But isn't the real moat relationships, not intelligence?
Yes - and this is the point most commentary misses. The biggest defence for incumbents isn't cognitive superiority. It's professional networks, client trust, and institutional access. A Magic Circle partner's value isn't only sharper legal reasoning. It's the ability to call the right person at the FCA, the trust built over twenty years of transactions, the reputation that gets you in the room. No API key opens the door to the boardroom.
The professionals who survive will be those who understand their value has two components: cognitive (judgment, pattern recognition, domain expertise) and structural (relationships, trust, access, reputation). AI compresses the first. It cannot touch the second.
Who trains the next generation?
This is the question nobody is answering. A trainee solicitor at a 200-person City firm currently spends her first three years doing due diligence reviews, drafting contract clauses, researching case law. That work is how she learns - not from the output, but from the thousands of hours of pattern exposure. Automate it, and you get a generation of brilliant seniors with no successors.
Medicine solved this with simulation. The military uses war games. The professions need their equivalent - structured mentorship, AI-assisted learning pathways, simulated case work - and they need it now, not in a decade when the gap has already opened [1].
What should you do this quarter?
Audit your team's work. Separate the cognitive output - research, drafting, modelling - from the judgment and relationship work. The first category is being repriced toward zero. The second is becoming more valuable. Restructure your teams accordingly: fewer juniors on production work, more investment in the client relationships and institutional knowledge that no model can replicate.
Then track the rung gap. After every major model release, ask: what percentage of our junior work could a frontier model now do at 80% quality? That number is your exposure.
The handloom weavers didn't vanish. They became factory workers, then machinists, then engineers. The transition took a generation. The professionals who thrive will be those who understand that their value was never in the loom.
If you're navigating this transition and want to think it through with someone who's been mapping the landscape, book a call with Lion Strategy.
Notes
[1] McKinsey Global Institute, "A New Future of Work: The Race to Deploy AI and Raise Skills in Europe and Beyond," May 2024. The one billion knowledge worker figure is an estimate of global workers whose tasks overlap with current LLM capabilities.