Not less. As AI takes on more of the routine cognitive work, the things only a human can bring become the things that matter most. →
The optimistic story is true. But it does not happen on its own. The default deployment of AI makes the human less visible, not more — because the metrics still reward throughput, and the AI optimises throughput faster than any person can.
And the engineering is not catching up: a 2026 Princeton study (Rabanser, Kapoor, Narayanan et al.) found reliability improving at one-seventh the rate of capability on the customer-service benchmark, and half the rate on the general agentic one. Capability is racing ahead of dependability.
For the promise to become real, four conditions have to be built deliberately.
Internal goods. The development of the practitioner over years, not units shipped this quarter. If the metric does not change, the AI will optimise the old metric and the human will disappear into it.
Because skipping them produces no judgement. A junior who never had to do the work without the tool will never develop the standing to push back on the tool when the tool is wrong.
The practitioner owns her own development, her own data, her own criterion of what counts as good. The twin reports to the practitioner — not the other way around.
Simulations push back hard — they do not say "good job." Coaching stays human where relational depth matters. Some difficulty is the medium through which competence forms. The whole game is to keep that difficulty intact.
The default climbs only Ladder 2 while the first is taken away. The third position climbs both together — each developing the other.
1. Are we deploying AI in a way that makes our people more capable or less, measured over years?
2. Do our juniors have a stage in the career path that AI cannot let them skip?
3. Does our coaching system concentrate human time on what only humans can do?
4. Does our data give the practitioner authorship — or the organisation surveillance?
5. Do our simulations push back hard enough to build judgement?
The human becoming
more visible
is the destination.
The path is in our hands.
Twin Ladder is the open standard (CC BY-SA 4.0) for measuring AI competence at the deployment layer, mapped to Article 4 of the EU AI Act. Seven conditions become a measurement framework. The brochure shows what to do on Monday.