For educators and institutions: what must change

Two and a half million graduates a year, prepared by a system that was built to produce the worker AI replaces fastest. The system did its job for a long time. Here is what would have to change, and why most of the change will not come from inside it.

Planned for release phase 2. The page is complete as a draft; interactive parts run in the browser and do not yet save to an account.

In brief

Three changes. What the system measures: marks are the shadow of capacity and the shadow can now be produced without it; measure capacity directly through projects and portfolios a student can defend. What it teaches: information is no longer scarce; the ability to think clearly with information, direct powerful tools and make judgement calls is. Who it trusts: certification is losing its place in the trust hierarchy to direct evidence, and institutions need to produce evidence, not stamps. None of these ideas is new; institutions move at a generational pace, so the students in the system now will mostly be saved by people three years older than them.

The worker the system was built to produce

Most schools and colleges in India still measure what they measured in 1996: marks, rote retention, performance on a fixed-format examination. Even then the system was optimising for a particular worker — one who could be reliably plugged into a hierarchy and do well-defined work to a well-defined standard. That worker was what the post-liberalisation economy needed, and the IT services industry was built on the back of them. The system delivered.

The problem is that the well-defined-work-to-a-well-defined-standard worker is exactly the worker being automated first, and the system is still producing them at scale. Most of them have no idea the world they are being prepared for is contracting, because the people preparing them did not navigate the contraction themselves and have no language for what is happening to their students.

Change one: what you measure

The grade is a shadow of capacity. For a long time the shadow was an adequate signal. It is not any more, because the shadow can be produced without the capacity. Measure capacity directly: projects, portfolios, structured work a student has built and can walk through line by line. A student should leave a four-year programme not with a transcript but with five things they can show another human being and defend. The transcript can be supplementary. The portfolio is the product.

Change two: what you teach

Information transfer was a defensible model when information was scarce. It is not scarce. What is scarce is the ability to think clearly with information, to direct powerful tools, to make judgement calls where judgement cannot be formula-ed. Teach what is now scarce. The framework is one version of that teaching; there are others. The point is that most institutions have not yet asked what their teaching is now actually for.

Change three: who you trust

The institutional model assumes the institution is the trusted authority: complete what it asks, exit with its certification, and the world treats you as competent. The certification can now be produced without the underlying capacity, so it means less, and the institution is slowly losing its place in the trust hierarchy to direct evidence — a project, a piece of work. Institutions need to start producing direct evidence, not just stamps.

Why it will be slow, and where the faster reform is

None of this is new; all three have been written about for a decade. They are not happening at scale because institutions move at a generational pace, and by the time most update their model, two or three cohorts will have been failed. The students in the system now are mostly not going to be saved by a reform. They are going to be saved, if at all, by people three years older than them who have been through the new world and can say how to navigate it — a cousin on the phone, a brother helping with a festival. That reform is faster, it can begin now, and it needs no institution's permission. "It doesn't announce itself. It feels like a phone call with your sister."

What an educator can do this term

Bring this to a college or a school.

Seminars for students and faculty, and courses built to run inside a term.

Questions people ask

Should colleges ban AI tools?
Banning produces Avoiders and drives use underground. The book's argument is to change what is measured — capacity, not its shadow — so that using the tool to skip learning stops paying, while using it to learn faster is taught explicitly.
How can a teacher tell who understood and who outsourced?
Ask them to defend it. A short oral walk-through of a piece of work separates capacity from output in minutes, which is exactly why senior architects in the book use it.
Is this only about engineering education?
No. The framework is shown working in an art college in Chapter 11, and the three changes apply to any programme whose grades can now be produced by a tool.