Economists from the American national Reserve (Fed Branch in Saint Louis) have just published a very interesting study. It was about measuring the impact of adaptation of artificial intelligence technology on the American labour market. Not theoretically, but "here and now" that is, between 2022 and 2025. This time scope gives us assurance that we are already talking about the planet of the first boom on the AI, which is already after the introduction of the first versions of the GPT hut into universal use. After which more followers came, they were constantly improved in the meantime.
What do you see?
There is simply a reasonably clear correlation between the level of manufacture vulnerability to AI and the increase in unemployment in this profession.
For example, the IT and math manufacture experienced an increase of 1.2 percent points in unemployment in the US from 2022 to 2025. At the same time, it is simply a profession with high, due to the fact that 80 percent vulnerability to AI technology.
Other examples of the same phenomenon are:
- business and finance (1.5 pp increase in unemployment and 60% vulnerability to artificial intelligence) and
- architecture and engineering (1.4 pp up erstwhile it comes to unemployment and 50% vulnerability to AI).
And on the another side of the mirror? The sectors that have mostly not experienced major changes in employment levels in fresh years are education, office administration and legal and social services. Everywhere, the AI was much smaller than in the economical areas described above (between 30 and 40%). The exception was the office administration, where despite a reasonably advanced vulnerability to AI (60%) unemployment did not grow.
Looking at this data, AI seems beautiful predictable. It's eating jobs right where it's expected to eat. The question is, of course, what happens next?
Will there be areas specified as management, education and rehabilitation or legal services? After all, everywhere we can actually imagine customers giving up the services of a live trainer, educator or legal advisor and alternatively they control to an AI model that gives a sense of equivalent service. Management besides fits in well. Why buy specified a service on the marketplace erstwhile it can easy be transferred to the GPT hut?
Robotisation and artificial intelligence
Recently, there was another – and rather fresh – work on a akin subject.
She wrote it. Florence Jaccoud from UNU-Merit, or the joint academic venture of the UN and the University of the Dutch Maastricht. The economist shows that
When you examine the impact of robotization (i.e. the earlier phase of the automation revolution) on wage equality and then compare it with the analogous relation between wages and the spread of artificial intelligence technology, we are in 2 different worlds.
Because robotics – in rule – reduced wage differences, especially among little earners. Meanwhile, artificial intelligence works precisely the opposite. Wherever algorithms or deep learning begin, the distribution between wages is intensified. This is especially evident on the top, which is the best earners. At least this is the consequence of a survey covering 19 European Union countries in the last 15 years. These 15 years are evidently to capture both phases of automation. And the robotization of 2010-2020 and boom of AI, which has been going on since 2022.
Hard and without anesthesia
– this is how we have been operating since 2020. Journalism that is not indifferent. The civilian Affairs Weekly announces abuse, educates and gives tools for a real, civic change.
Donate and become our contributor
So we have 2 generations of automation. This somewhat older, or robotization, involves introducing more machines into production processes.
These machines replace a man in more hard tasks, which he is incapable to do physically or would take him a lot more time to perform. And we have artificial intelligence that replaces a man with a akin scheme, but in the field of intellectual work – in graphics, programming, writing or sweeping immense databases. The apparent effects of both automation should be similar. But they're not. In fact, these consequences go in precisely other directions. Why? Florence Jaccoudi's explanation is that robotization is complementary to regular physical tasks. Meanwhile, AI technologies importantly increase the productivity of highly specialized cognitive work.
Workers: Together or separately?
However, I would alternatively search answers in an organization environment. In manufacture and classical manufacturing robotization appears in places where the planet of work is more self-conscious and organized (although in trade unions). As a result, the workers' planet has more strength to request that the productivity gains that robotization brings be distributed more reasonably among workers. That would explain the lower pay inequalities.
On the another hand, in the labour marketplace segment, where AI is raging, the situation looks very different. It's full of workers operating according to the rule of "everyone's kneecap starch". As a result, the increase in productivity goes entirely to those at the end of the value-added chain. The intermediate cells stay behind. Or they're out of the game at all. Which would explain the increase in wage inequality and the reports of artificial intelligence technologies gnawing at the well-being of a simple programmer who was inactive in bids yesterday, and present he's worried about tomorrow.
Either way, both of these works – and this Fed from Saint Louis and Mrs Jaccoudi – prove the same (in their own way). Artificial intelligence eats the work and makes our planet little equal. That's just the facts.
We're, of course, at the beginning of this trial. However, concerns are considerable. They even sounded at an economical Nobel Prize convention in the charming Bavarian Lindau organized a fewer months ago. On technological unemployment they clashed there Simon Johnson (Nobel 2024) and Christopher Pissarides (Nobel 2010).
– The mountain copes, the bottom struggles for survival, the mediate is squeezed very hard. [...] And if you lose your position inside, you are pushed to the very bottom of the labour market, where it is very hard to make a surviving – Johnson said during his lecture. In a akin speech his last sound is besides heard A book written to a company with Daron Acemoglu, about which I wrote in the civilian Affairs Week A while ago.
A little alarming message was made in Lindau by Christopher A. Pissarides. In his opinion, concerns about artificial intelligence are exaggerated by panic commentators and upsetting media. – There are presently no employees, not jobs. There is any incompatibility in the economy. However, given the decline in the population, the ageing population, the deficiency of migration, etc., a shortage of workers, alternatively than jobs, is expected, he argued.
However, in this dispute, I take Johnson's side. besides due to the fact that we are beginning to see this not only in the anecdotal stories of life taken. For example, from stories that have been experiencing a burst bubble in the IT marketplace of employees over the years of the ever-growing sector. Now we see this besides on circumstantial numbers. And I feel like there's gonna be more.
