Will AI make humans redundant?

First published 17 May 2026

What we'll be valued for instead.

AI hasn't changed the tool. It has changed what institutions believe good looks like. And that change lands on everyone.

Not redundant. Re-priced. The World Economic Forum expects 39% of the skills employers value to change by 2030, and PwC finds the requirements of AI-exposed jobs changing 66% faster than the rest, with a 56% pay premium for people who can work the tools. AI hasn't changed what you can do. It has changed what institutions believe good looks like, and that lands on everyone, including people who never opened a chatbot.

Every mass technology finds the human weakness it can most profitably exploit. Social media found status anxiety — built a $1.4 trillion advertising industry on the gap between who you are and who others think you are — and restructured the social contract around it. You didn't have to post to feel the consequences. You just had to live in a world that had reorganised its definition of relevance around a metric you never chose. AI is running the same play. Not on your sense of status relative to other people. On your sense of value relative to the machine. And the rewrite is already underway.

There are two AI stories being told right now. The first is personal — what AI does to the individual who uses it, the cognitive shortcuts, the atrophied judgment, the slow erosion of the ability to think without assistance. That story is real and it is covered elsewhere. The second story is less visible and more structurally dangerous: what happens when AI adoption at scale rewrites what institutions believe good looks like — and that rewrite lands on everyone, including the people who never opted in.

David teaches English in Stoke. His year eleven students submit essays that are structurally perfect and completely unreadable as evidence of understanding. He can't prove they didn't write them. The exam board hasn't updated its rubric. He gives them the marks. Nothing has broken. The metric has just decoupled from the thing it was built to measure, invisibly, and the institution has started rewarding the output of the machine as though it were evidence of the thinking that used to produce it. David didn't choose this. His students didn't choose this. The rubric changed around them.

The people most at risk are the ones whose value was always in the slower, harder-to-measure, irreducibly human work — and who now find themselves evaluated against a metric that was never built to see them. The homogenisation of what good looks like is already underway.

Below: here and now, imminent, on the horizon, the long view, and five behavioural shifts running underneath.

01. Here and Now

The metric has already changed. Most institutions haven't noticed yet.

The decoupling of output from judgment is not coming. It is here. Hiring managers, exam boards, and commissioning editors are already making decisions against a rubric that AI has quietly invalidated. The individuals being passed over mostly don't know why.

• The rubric breaks silently — When AI produces essays, briefs, and reports indistinguishable from human work, the instruments built to identify human thinking stop working. No announcement is made. The standard just drifts.
  • Exam boards mark what they can no longer verify — David's marking scheme rewards structure, argument, and citation. AI delivers all three without understanding any of them. The grade goes up. The evidence of learning disappears.
  • Hiring processes reward the wrong signal — Cover letters, take-home tasks, and written assessments were proxies for judgment and communication. They are now proxies for prompt literacy. The underlying capability being selected for has changed without the job description changing.
• Speed becomes the floor, not the differentiator — AI makes fast, polished output the baseline expectation. The person who needs three days to produce something genuinely considered doesn't get credit for the consideration. They get penalised for the three days.
  • The considered thinker reads as slow — In a world where a confident, well-structured answer arrives in minutes, the person who says 'I'd need to look into this before I had a view' sounds unprepared. Same epistemic honesty. Completely different social reward.
  • Some industries haven't rewritten the rubric yet — Law, medicine, and academia still formally reward demonstrated reasoning. The pressure is building but the instruments haven't broken yet. This is a stay of execution, not immunity.
• Individuals who never adopted AI feel the first effects — The person who never used AI is already being evaluated against outputs produced by people who did. The comparison is invisible and the gap is widening.

02. Imminent

Institutions start optimising for the new metric. The ones being devalued start to notice.

Within the next six to twelve months, the institutions that have drifted toward speed and volume as proxies for competence will begin to formalise that drift. Job descriptions will change. Assessment frameworks will be rewritten. The people whose value lived in the old framework will find the ground has shifted.

• Job descriptions are rewritten around AI fluency — Roles that previously required demonstrated judgment now list AI tool proficiency as a core competency. The underlying capability being hired for has changed. The salary band hasn't.
  • Senior roles hollowed out from below — Junior AI-fluent hires produce outputs that previously required five years of experience. The mid-career professional whose value was in accumulated judgment finds their position harder to justify at their price point.
• Creative industries reprice human originality downward — When AI can produce competent copy, design, and music at volume, the market rate for competent human work collapses. The individuals who produce genuinely original work cannot prove it is original at the speed the market now expects.
  • Homogenisation accelerates — Commissioning editors, creative directors, and brand managers under output pressure default to AI-generated work that is good enough. The range of voices in circulation narrows. Nobody makes the decision to narrow it.
  • A premium market for verified human work emerges — Some buyers begin paying explicitly for provably human output. The market bifurcates. The premium tier is small and the volume tier is vast.
• Institutions begin to notice what they have lost — Early adopters of AI-first workflows start producing work that is fast, consistent, and indistinct. Some notice the problem. Most do not have the instrument to diagnose it.

03. On the Horizon

The social contract rewrites itself. The new definition of competence becomes the only definition.

Within one to three years, the metric shift moves from institutional drift to formal standard. Assessment frameworks, professional accreditation, and hiring practice codify the new definition of good. The window for the old framework to reassert itself closes.

• Professional accreditation catches up with the wrong standard — Bar exams, medical licensing, and professional certifications are rewritten to reflect AI-assisted practice. The revised standard measures competence at working with AI, not competence at the underlying discipline.
  • The next generation is trained for the proxy, not the thing — Students optimise for the assessment. The assessment now rewards AI-fluent output. The foundational reasoning the assessment was built to develop stops being taught because it stops being tested.
• The monoculture becomes visible — Across media, professional services, and education, the range of outputs in circulation has narrowed to what AI produces well. The narrowing is only visible in retrospect, when the thing that has gone is no longer there to compare against.
• Regulation attempts to draw the line — Governments and professional bodies mandate disclosure of AI use in assessed work. The mandate is difficult to enforce and easy to route around. It signals concern without solving the structural problem.
• Some institutions redesign assessment from the ground up — A minority of schools, universities, and employers rebuild their evaluation frameworks around capabilities AI cannot replicate — live reasoning, verbal defence, contextual judgment. These institutions become talent magnets for the people the market has stopped seeing.

04. The Long View

The dust settles on a world that optimised for the wrong thing.

Three to ten years out, the structural consequences of rewriting institutional value around AI-speed competence become legible. The individuals who were devalued in the transition either found new frameworks that could see them, or they didn't. The institutions that lost the ability to identify human judgment find out what that costs.

• Institutional brittleness compounds — Organisations that replaced judgment with AI-speed output find themselves unable to navigate situations that require genuine reasoning under uncertainty. The capability was priced out before anyone noticed it was load-bearing.
  • Crisis performance collapses — The situations that most require human judgment — ethical dilemmas, novel problems, high-stakes decisions with incomplete information — are precisely the situations AI handles worst. Institutions that have atrophied their judgment capacity find out at the worst possible moment.
• A generation of professionals cannot function without the machine — The cohort trained entirely within AI-first frameworks reaches senior positions without having developed the foundational reasoning those positions require. The gap between their confidence and their capability is wide and invisible to them.
• The social contract stabilises around the new normal — As with social media, most people adjust to the new definition of value without consciously accepting it. The adjustment looks like adaptation. It is also loss.
• Human judgment becomes genuinely scarce and genuinely valued — In a market flooded with AI output, the ability to reason independently, tolerate uncertainty, and arrive at a defensible position becomes rare enough to command a real premium. The individuals who held onto the capability — and found institutions that could see it — are in short supply.

05. The Behavioural Rabbit Hole

The shifts the time arc doesn't capture.

These are the behavioural changes running underneath the institutional story — the ways that hiring managers, educators, commissioners, and peers are already changing what they reward and how they see value, without any formal decision being made.

• Hiring managers stop being able to read the signals they used to trust — The CV, the cover letter, the written assessment — all of them used to surface something about how a candidate thinks. Hiring managers who know this know they can no longer trust it. Most are proceeding as though the instruments still work.
• The meeting room rewards the wrong person — The person who runs a question through AI before the meeting and arrives with a confident answer gets the follow-up. The person who says they would need more time to form a view gets passed over. The organisation learns to prefer performed certainty over honest uncertainty, one meeting at a time.
• Commissioners and editors stop commissioning what they cannot benchmark — When AI produces competent work at volume, the editor under deadline pressure stops commissioning the writer who takes longer and is harder to predict. The work that gets made narrows to what can be evaluated quickly against a legible standard.
• Peer recognition shifts toward output volume — In professional communities, the people producing most visibly and most frequently gain status. The people doing slower, less frequent, harder-to-categorise work become less legible to their peers. The community's definition of who is good at this drifts without anyone voting on it.
• Some managers actively redesign for what AI cannot do — A minority of leaders — having watched their teams produce fast, indistinct, undefendable work — start deliberately creating conditions for slower thinking. Longer briefs. Verbal presentations instead of decks. Questions with no right answer. They are swimming against the current and they know it.