Most AI job plans are for last month's AI

The Mayor of London's AI and Jobs Taskforce published its report today: Shaping how AI changes work in London. Chaired by Baroness Martha Lane-Fox, with John Amaechi as the only psychologist on the group.

Its best line is also the right starting point: AI is inevitable. How work changes because of it is not.

I agree with that. I also think many people - and many careful plans - are only considering the AI of a few months ago.

What the report gets right

The Taskforce does not buy the idea that all jobs vanish overnight. That matches what the data has been saying. The loud story is mass job loss. The quieter story is fewer first jobs, harder promotion paths, and change done to people rather than with them.

GLA numbers in the report put about 2.4 million Londoners in roles where generative AI could take over a large part of their tasks, with more than a million in highly exposed jobs. London will feel this early because of how its economy is built - professional services, finance, creative work, tech.

Their response is practical:

• Act early - better, faster information on hiring and skills so partners can respond before graduate hiring falls or shortages become fixed.
• Shape change - a partnership where employers invest in people as well as tools, redesign work, and give workers a real say.
• Create opportunity - a London Future Ready Guarantee for skills that matter more as routine work moves to machines: judgement, communication, adaptability, ethics, and the rest of the list that exams barely capture. Plus new routes in when old starter jobs shrink.

They also note freelancing, mixed careers, and small businesses rising - helping people not only find work but create it. For some people that will be true. It is not a general answer for everyone.

If you work in skills, HR, policy, or education, read the full report. It is better than most of the headline fight between utopia and doom.

Where the plan still looks short

Institutions plan on yearly reviews, long buying cycles, and skills programmes that take months to redesign. AI models do not wait for that clock.

Most of what we still call "AI at work" is today's chat and writing tools doing more of the language, analysis, and admin that junior and mid-level office work used to own. That is real, and the report is strong on it.

The next versions of these systems will be better than the ones firms are trying this quarter. Not in a sci-fi way - in the boring way that coding, research, customer work, and planning keep getting faster and cheaper. A plan built around last winter's tools is already late.

And office work is only half the story. Humanoid robots and other machines that do physical work are starting to look serious for warehouses, logistics, some care and service jobs, and factories. How fast is still debated. The direction is not. A London report that focuses on generative AI and office careers can still be strong and underplay a second wave: fear of becoming obsolete for people whose value has been physical skill, stamina, and being present.

So the risk is not that City Hall is wrong about entry-level professional work. The risk is that we plan well for the disruption we can already see, while the next jump in what AI can do is already being trained.

Not everyone can "just freelance"

This is the part I wish more AI-and-jobs writing would face.

Mixed careers, creative pivots, and "create your own work" sound hopeful, but they are a poor answer for someone who drove a truck for thirty years, or stacked shelves on a rota that paid the rent, or did solid admin that was never glamorous and never needed a personal brand.

Even on a simple Big Five cut the point lands. People high in conscientiousness - and many who are lower in openness to experience - tend to want solid, reliable work: clear duties, structure, a job that stays put. That is a large share of any population, not a niche. Plenty of people simply do not want freelancing, constant reinvention, or selling themselves as a product. That preference is not a character flaw. It is a common human pattern.

And it is exactly the kind of work automation is aimed at: predictable routes, repeatable tasks, steady hours, a role you can master and keep. When that goes, "become a creative freelancer" is not a path. It is the wrong answer for the problem.

So when reports celebrate new forms of work, they should also say plainly: some people will not successfully move into the jobs the planners prefer. Knowing yourself helps you see what kind of work fits you. It does not invent trucking jobs if the trucks drive themselves.

There is also a generational pattern that plans often blur - a tendency, not a rule. A new generation of young people will grow up inside the new world and many of them will adapt - new tools, mixed careers, work that looks strange to their parents. People who spent decades building skill and identity around a solid job often find the jump harder. That is not stubbornness as a moral failing, and it is not true that no lorry driver can adapt or that only the young can. It is how long careers and human identity often work. Skills programmes and "Future Ready" badges help some mid-career people. They do not turn every long-serving driver into a portfolio creative on demand. Policy that only designs for the most adaptable will leave a lot of people behind.

What if whole groups of jobs go?

The Taskforce is careful not to promise mass unemployment tomorrow. Fair enough. The harder official question is still underplayed:

What happens if, over a decade, whole groups of jobs are not "redesigned" but largely replaced - office work first, then physical work - faster than good new work appears for the people who held the old roles?

That is not a sci-fi movie. It is a step-by-step picture: fewer junior analysts, then fewer routine drivers and warehouse roles, then the next layer. At each step the people left need income, status, and a story of contribution. Retraining helps some. Partnerships help some. Better jobs data helps some.

At some scale you are not only talking about skills programmes. You are talking about the social safety net: income support, housing, health, local shock when a sector shrinks, dignity without a traditional job title. Welfare, tax, and national employment law sit largely with national government - the report says as much. Still: has this been considered seriously enough in official AI-and-work work, or do we stop at apprenticeships and "Future Ready" badges because full replacement of whole job groups is hard to talk about in public?

If London (or the UK) only plans for task change and career pivots, and what AI can do keeps rising, we will find the safety-net gap the hard way. Reports that want to shape how AI changes work should start asking, out loud, what progressive replacement of whole job families needs from the state - not only from employers and colleges.

What that means for a person

The Taskforce is right that redesigning work is a choice. Employers decide what to automate, what stays with people, how gains are shared, and whether new routes into work open when old ones close. Individuals do not control those choices. We do control how early we get honest about our own position - including whether we thrive on freelance uncertainty or need solid, reliable work.

"Future Ready" skills - judgement, collaboration, learning speed, ethical sense - are not abstract virtues. They sit on top of something more basic: knowing how you work under pressure, what you are good at when the job title wobbles, how you respond when the ladder moves, and who you are when the old story about your career stops fitting.

That is the FOBO point I keep coming back to - fear of becoming obsolete. It is partly about money. It is also about identity. When a tool does the thing you were paid to be good at, the first crack is often not the payslip. It is the story of who you are.

Retraining and apprenticeships matter. So does worker voice in how AI is rolled out. So does early jobs data. So does an honest national conversation about income and dignity if whole groups of jobs shrink faster than replacements appear. None of that replaces knowing yourself as the portable bit you take with you - and none of it makes self-knowledge a stand-in for a safety net when the work simply is not there.

Not a choice between tech and people

I am not arguing we slow the tech. I am arguing we stop pretending that a plan built on this year's tools, plus freelancing for the few who can handle it, is a plan for the decade.

London can lead by acting early and redesigning work with people in the room. Firms can invest in people as well as tools. Educators can stop treating adaptive skills as soft decoration. National government has to own the question of what happens when replacement, not redesign, becomes the pattern in more and more areas of work.

And individuals can treat the next few years as a window: still uneven, still mostly people doing the work in many places, still time to map what you bring - and to press institutions to plan for people who will never be freelancers.

The report is a solid start for the city but we need to plan one step further than the AI you already know, and one step further than "everyone reinvents themselves."

Our free portraits on what-world-way were built for personal insight in times of change, drawing on stage and adult-development ideas in psychology. This will not be the first mass social change - but it may reach further, and sooner, than most. We are a non-profit; our hope is that the portraits get used widely enough to help a little, where self-knowledge is useful. They do not replace jobs, training, or a safety net.