will AI put executives out of work? (Denis Jacquet facing Sarah Guillou)

2023-06-01 04:03:00

There are many concerns raised by the emergence of ChatGPT. Even more than the spread of false information or the manipulation of democracies, the world of work fears the upheavals of generative AI. A study by the University of Pennsylvania and OpenAI, the company behind ChatGPT, estimates that the ChatGPT-4 version translates text, writes, processes data or produces lines of code twice as fast as a competent person. . Microsoft, OpenAI investor, also notes an identical level of performance between the ChatGPT4 machine and humans in medicine, law or mathematics.

For or against: should access to the RSA be conditional? (Karl Olive against Jean-Claude Barbier)

This time, the losers of technological progress could not be workers but white collar workers, starting with accountants, auditors, translators or mathematicians, according to the University of Pennsylvania. In total, Goldman Sachs estimates that 34% of executives and skilled professions are exposed to automation through AI. So, will AI put executives out of work?

Between those who sweep aside the alarmist arguments with a distracted gesture and those who announce the domination of the Matrix, the debate is a little light. Despite the incantations of big boss of Silicon Valley, we will not be able to regulate AI. You can’t stop the sand that slips through your hands. There will not be global unanimity on the subject, but without China, there can be no domesticated AI. So we are going to suffer, in two ways.

Automation will affect all “white collar” jobs, which is why everyone is panicking. As long as we took the jobs of the poor, via robotization, no one cared. A poor man who is a little poorer does not pity anyone. But there we touch those “from the top” and they don’t like it. Automation will perform, better and faster, what they have been doing until now. This will inevitably reduce the space available for humans. We will not immediately kill millions of jobs, but we will create fewer of them. This is already the case in a thousand areas.

However, there will be a dampener through demographics, as seniors will continue to leave the company, without it being able to find enough young people to replace them. There will thus be fewer jobs, but also fewer young people to take them, creating a form of balance, but far below what is necessary. We will never be able to create as many jobs as those that automation and robotization will destroy. More demography means less growth. So fewer jobs. Gained productivity will not turn into new jobs. We have to be realistic.

For or against: should the development of artificial intelligence be slowed down? (Colin de la Higuera against Paul Midy)

But much more serious still, it will decrease the value attributed to the work of white-collar workers, which will impoverish a middle and upper class. We need it because it is the backbone of our societies, which will thus continue to slide downwards, which will complete the impoverishment of an already largely weakened middle class in the West. Why pay such an expensive auditor when the AI ​​is already doing 80% of its work? Why pay a radiologist €800 for his act, when the AI ​​does better than him? They will be paid less.

All jobs can be automated from 42% to 82%. “The homo Colus Blancus”, who collects data, analyzes it and draws conclusions, will no longer be necessary. His place will be reduced, his salary too. This is a much more worrying subject than the disappearance of jobs, because it will come much faster.

During this time countries with negative demographics, such as China, Russia, Japan, will invent and perfect techniques capable of replacing humans, out of necessity. Not enough arms and brains. They must therefore be replaced by AIs. This will boost their margins, and will require us to do the same to remain competitive. In a second step, we will therefore have to lay off workers to remain competitive, and this is where the job-killing machine will massively come into play.

One can be optimistic by nature, but to remain blind is unconscious. We must prepare now to change the economic model of our society and our businesses. Otherwise it is a “carnage” guaranteed to less than 15 years.

Against

Two fears dominate the acceleration of the potentialities of Artificial Intelligence (AI): that of malicious and anti-democratic use, that of the disappearance of jobs. If the first seems to be of increasing serious concern to AI players themselves and experts in international relations, the second worries individuals more directly because it threatens their main source of income. The fear of the effects of technical progress on employment is not new, but AI has deeply disruptive characteristics and will spread to all professions, even executive professions.

AI is a general-purpose technology with great potential for duplicating human intelligence. This technology is therefore intended to replace hours of human work. Is it different from information and communication technologies and robotization? In 2013, the publication of two Oxford researchers, Frey and Osborne, was very alarmist. They announced that 40% of jobs were threatened by automation.

Ten years later and despite a major shock to the economy (pandemic), the tension on certain labor markets shows more the shortage of workers than the great downgrading that was predicted. Ten years later, the anxiety is back again, exemplified by a Goldman Sachs report released in April 2023 that still presents a gloomy outlook, this time in response to AI penetration. They announce that 75 % of US industrial employment would be altered by AI and that a quarter of them will experience an alteration that could lead to substitution.

For or against: should we reconnect with budgetary rigor and reduce state spending? (Jean-Marc Daniel against Thibault Laurentjoye)

But being affected by AI doesn’t mean jobs will go away. The extent of the change depends on the sequence of tasks that make up the job. The effect of AI on jobs depends on the intensity of penetration into the job concerned. As long as there remains a residual task that must be carried out by humans, then employment will gain in productivity but will not disappear. Everything therefore depends on the interweaving of the succession of tasks, and on the need for human control or arbitration at one point in this network of interwoven tasks.

AI will first be instrumental and increase the value, quality and margins of the products and services of those who will adopt it. In doing so, it is very possible that an accounting firm will part with part of its employees because the AI ​​will replace the execution of their work tasks. However, the accounting firm could increase its market share and the quality of its services by keeping its employees deployed on residual human tasks, the number of which has increased with its market share.

It is also possible, as in all cycles of technological advances, that new job opportunities associated with AI will appear and in particular because AI will increase human capacities to innovate. Jobs will disappear, jobs will change and new jobs will appear. The question is to know at what rate the reallocation of work between past jobs and new jobs will take place. Often it is the conditions of production that are decisive: a market undergoing a negative productivity shock (problem of resources for example) adapts by adopting new technologies. AI will be adopted if it is an answer to a cost problem.