While fears are regularly expressed about the threat that AI would pose to employment, a report by the International Labor Organization paints a less pessimistic picture. Employees will not be massively laid off because of generative AI.
Artificial intelligence, gravedigger of the jobs occupied by humans? This is the recurrent fear that is expressed with each development of the sector. The arrival of generative AI, symbolized by projects like ChatGPT (text) and Midjourney (image), has obviously fueled this concern: the algorithms are becoming so efficient that they could eliminate jobs.
A report by the International Labor Organization (ILO) takes the opposite view of the Cassandres, anticipating not a destruction of jobs with the emergence of generative AI, but rather an increase in creation. The conclusions of the study can be read in the document Generative AI and employment: a global analysis of the potential effects on the quantity and quality of employment.
The report commissioned by the ILO, a specialized agency of the United Nations (UN), finds that ” most jobs and industries are only partially exposed to automation “. Under these conditions, failing to be able to achieve complete automation, we will instead see the use of generative AI in support of existing positions.
Accordingly, the ILO anticipates “ potential changes in job quality, including work intensity and autonomy.” These generative tools will serve as virtual assistants or trainees: we can delegate repetitive, time-consuming or daunting tasks to them, so as to focus on activities with higher added value.
An analogy is often made with Excel software and accountants: the development of spreadsheets has not made this profession disappear. This one seized it and Excel offered productivity gains to accountants, and time to take care of more complex missions. Observers believe this same scenario will play out with generative AI.
This optimistic reading of the future is however nuanced by the ILO, since its report admits that some positions are more exposed than others. Generative AI is going to have a very strong impact on basic clerical jobs, with a quarter of jobs considered high exposure and more than half of jobs having a medium level of exposure.
For the other professional categories, exposure to AI plummets: technicians and intermediate professions have moderate exposure for a quarter of their tasks. Managers for just over 10% of their activities. The other businesses present practically negligible exposure. Almost none of them have significant strong exposure.
An analysis limited to generative AI
It should be noted that the ILO report, shared on August 21, only covers the generative AI sector. The reflection of the study does not unfold on a broader scenario, with the so-called general AI, which does not exist today. It only looks at the existing and the foreseeable on a reasonable horizon, without assuming a strategic surprise that would reshuffle the cards.
It is also a report to be distinguished from the work of robots, which are programmed to perform tasks without necessarily being backed by AI. Here, the estimates are more pessimistic: in 2017, the McKinsey Institute estimated that 40% of jobs in France can be replaced by machines. Again, the prediction depends on the job being considered, and not everyone agrees.
As it stands, the report assumes that 5.5% of total employment in high-income countries (like France) is potentially exposed to the automation effects of technology. In the case of low-income countries, this rate drops to 0.4% of all jobs. We are very far from the apocalypse sometimes predicted.
However, the ILO issues a warning concerning women: they are potentially more exposed to automation than men, because they are traditionally much more present in office jobs. The report calls on public authorities to take this into account, as women could be more affected by generative AI.
Beyond existing positions, generative AI could also be a source of job creation — for example, technicians and engineers specialized in designing good prompts to generate adequate results, both in text and visual form. . There will also be other professions whose existence we do not yet suspect today.
” 20 years ago there were no community managers, 30 years ago there were few web designers and no amount of data modeling would have allowed making a priori predictions about a wide range other professions that have emerged in recent decades “recalls the study. Just as we never thought that there would ever be web influencers.
Do you want to know everything about the mobility of tomorrow, from electric cars to pedelecs? Subscribe now to our Watt Else newsletter!