Cognitive automation hits the entry-level jobs

Discussions of automation repeat a reassuring line: it has happened before, it always worked out, work recomposed. That is true, and it is also why it is worth looking closely at where this wave resembles the earlier ones and where it does not.

The resemblance is that this time too, occupations we cannot yet imagine will be created. The difference, the one that should worry us, concerns where on the professional ladder the automation enters.

Where machines entered, where software enters

Industrial automation mostly replaced repetitive manual work: assembling, welding, moving, packing. People who lost those jobs lost an occupation, often a well-paid one, and the recomposition was painful and slow. But entry into working life remained possible elsewhere: in services, in offices, in the professions.

Cognitive automation works on different material. It does not touch physical work: it touches the first draft, the summary of a long document, the translation, the classification of files, the preliminary research, the filling in of a template, the consistency check on a text.

List those tasks and then ask who performs them in an organisation. Not the senior people, not the experienced professionals. They are performed by the people who have just arrived.

The ladder that loses its first rungs

Here is the point that makes this transition different from its predecessors.

In almost every skilled occupation, learning happens through a period of doing simple things under the supervision of someone who knows how. The trainee lawyer drafting first submissions, the junior analyst gathering data, the assistant transcribing meetings, the new editor rewriting press releases.

Those are low-value tasks, and that is exactly why they were given to people who needed to learn. The value to the organisation was modest; the value to the person was enormous, because that is how a trade is absorbed — by making mistakes on things that do no damage, while someone watches.

If those tasks are performed by an automated system in a tenth of the time, the organisation saves money. But it also loses the mechanism by which it trained its own future professionals, and the person loses the rung they climbed from.

It is a problem people talk about little, because its effects are not immediate. A law firm that stops taking trainees has no problem this year: it will have one in ten years, when the current partners retire and nobody has been trained to replace them.

Why «retrain» is not enough

The standard answer to any discussion of automation is training: people retrain and move to the new occupations.

It is right in the abstract and insufficient in practice, for three reasons.

The first is time. Retraining seriously takes months or years, during which you have to live. Someone with a mortgage and two children cannot afford a year of training on no income, and this holds all the more for someone who loses their job at fifty.

The second is direction. Retrain towards what? When the target moves every eighteen months, advising someone to train in a specific skill is risky advice. Many people who trained in recent years on tasks considered safe have found those tasks among the first to be automated.

The third is that entry-level positions matter to people retraining too. Someone changing occupation at forty generally re-enters from the bottom: they do the simple work of the new sector for a period. If that simple work no longer exists, retraining has nowhere to land.

The cost falling on the young

Every transition has a group that pays a disproportionate share of the bill. In the industrial transitions it was adult workers in the affected sectors, with social consequences Europe and the United States have lived with for decades.

In this one the exposed group is different: the people who have to get in now. Someone who is twenty, with a qualification just completed and no experience to put on a CV, arrives in a market where the tasks people used to start with are exactly the ones organisations have stopped paying for.

It produces a cruel effect: experience is demanded for access to positions that were the only way of acquiring it.

What basic income would change

The objection at this point is legitimate: a basic income creates no jobs and teaches nobody a trade. That is true, and claiming otherwise would be dishonest.

It does three things that matter in the picture described.

It makes training without income possible. A year of study or retraining, impossible today for anyone without savings, becomes a workable choice. Not easy: workable.

It makes badly paid entry sustainable. Many learning opportunities — internships, short collaborations, first commissions — are economically impossible for anyone without family support. That is why some professions are in practice reserved for people who can afford not to earn at the start. A floor beneath your feet reopens those doors to people whose parents are not paying the rent.

It makes discontinuous work sustainable. The market automation produces is made of short assignments and fixed-term projects. With a guaranteed income that is never withdrawn, every assignment is extra income rather than a risk to be assessed.

A historical precedent worth remembering

In early nineteenth-century England, the mechanisation of spinning destroyed the work of hundreds of thousands of handloom weavers within a few years. Recomposition came, but it took decades, and in the meantime an entire generation paid a price that long-run statistics hide.

Economists now call that period the «Engels pause»: a long phase in which productivity rose and real wages did not. The benefit of the technology existed; it simply did not reach the people bearing its costs, and took a generation to get there.

The point is not that history repeats exactly. It is that the reassuring argument — «in the end it recomposes» — is true on timescales that do not match a human life. Telling a forty-five-year-old that in twenty-five years the labour market will have rebalanced is, from where they stand, an answer without meaning.

Basic income does not accelerate the recomposition. It makes the interval habitable, and it is precisely in the interval that a society decides whether a technological transformation is experienced as progress or as an assault. The most violent political reactions to industrial transitions did not come from people who were worst off in absolute terms: they came from people who watched their position collapse while nobody attended to it.

What has to sit alongside it

It would be wrong to leave the impression that this is enough.

The underlying problem is that training new entrants has become a pure cost for the employer, borne now for a future benefit that might go to a competitor. That is the classic case where public intervention has a solid justification.

The instruments exist and are not new: hiring incentives for entrants, fiscal recognition of time spent on internal training, publicly supported apprenticeships, training routes that rebuild outside the firm what firms no longer teach.

None of those instruments works if the person meant to use them cannot support themselves meanwhile. That is the right way to understand basic income here: not the solution, but the condition that makes the solutions workable.

What firms could do, if they wanted to

Closing by assigning everything to public policy would be convenient and incomplete.

Organisations automating entry-level tasks have a direct interest — deferred but real — in not dismantling their own training mechanism. In ten years they will need experienced professionals, and those professionals are being formed now.

Some routes require no public intervention at all. Keeping entry positions even when the traditional tasks are automated, by changing their content: reviewing automated output, quality control, handling anomalous cases. Pairing new entrants with experienced staff from the start rather than after a period of grunt work. Treating training as an investment on the books rather than a free by-product.

These are choices that cost money in the short run, and anyone managing to the quarter will not make them. Which is precisely why, absent a public incentive, almost nobody will.

A note on urgency

There is a reason this problem deserves more attention than it gets, and it concerns irreversibility.

An adult who loses a job at fifty suffers serious harm bounded in time. A young person who cannot get into an occupation for five years does not recover those five years: they enter later, with less experience, in a worse position, and that worse position shapes the whole subsequent trajectory.

The long-run effects of entering the labour market in a bad period are documented: people who graduate into a recession earn less for years than people who graduate two years earlier or later, all else equal.

If that holds for a recession lasting a few years, it holds all the more for a structural change in entry-level tasks. It is the kind of damage that cannot be repaired afterwards.

What would have to be measured

The indicators by which Europe tracks its labour market — employment rate, unemployment rate, hours worked — were designed for a world in which the question was whether you had a job.

They do not capture what is changing: the stability of trajectories, the number of forced transitions, the distance between skills held and skills demanded, the share of income coming from discontinuous activity.

It is entirely possible that cognitive automation produces, over the coming years, a labour market with stable employment rates and far more unstable working lives. With current statistical instruments, that transformation would register as «no problem».

Anyone with public responsibilities should be worried about that now, rather than explaining in ten years why the data showed nothing while people were saying the opposite.

In brief

This article addresses one part of the problem, not the problem. A basic income creates no jobs, teaches no trades, does not fix pensions and does not replace the services that answer unequal needs. The claim argued on this site is narrower: that it is, among the available instruments, the one that makes the others workable. Training requires being able to live while training; changing occupation requires being able to be wrong; refusing indecent conditions requires being able to afford refusal.

Three objections remain standing, and this site gives each of them a page: the cost, which is real and requires serious fiscal choices; the effect on rents in markets with inelastic housing supply; and the absence of a credible transition plan. None of them shows the proposal is wrong. All three show it is not ready, which is why the demand made here is not immediate introduction but a serious European trial and a public study putting costs, routes and losers on the table.

If anything on this page is wrong or badly argued, telling us is the most useful contribution available: the address is on the Contact page, and reasoned corrections are published.