Artificial Intelligence, First Numbers in the World of Work


A strong debate is developing on artificial intelligence that also involves the world of work. A first study shows that almost all job positions will have to do with some form of AI.

There has been a lot of talk about artificial intelligence in recent weeks. The matter is not new, but the evolution of applications based on GPTs has certainly accelerated in recent times. Many know software such as ChatGPT and experts are already experimenting with the integration of these technologies into the search engines of Microsoft, Baidu, and Google.

Furthermore, the AI-generated photos of the arrest of former US president Donald Trump, very realistic at first sight, have caused a sensation, relaunching the debate on the dissemination of fake news through social networks.

Elon Musk, the owner of Twitter, announced a few days ago that the feed of the “For you” section of the blue application will soon be fed only by content from verified profiles. It is the only way, Musk argues, to protect oneself from the almost certain deluge of content produced by artificial intelligence that will invade social networks in the coming months. Furthermore, the latest news involves the publication of an appeal by the non-profit organisation Future of Life Institute asking to suspend the development of the latest generation of AI (the GPT-4) for a few months and to discuss the consequences of such technologies.

But the issue of the progressive use of artificial intelligence is also affecting – and worrying – millions of workers. If we assume – and there is a good chance it is – that the advent of these new technologies will eliminate jobs but also create new ones, then it is likely that we are approaching another major turning point for the world of work. Are the legislators, the companies, and the workers themselves aware of it? Are they preparing for this huge upheaval? Hard to tell. 

In the USA the subject is topical and this is demonstrated by the many studies that universities, research institutes, and the companies themselves that develop this technology are publishing. The University of Pennsylvania and OpenAI (the company that created ChatGPT) published a very interesting analysis a couple of days ago. The working paper tries to draw up a first, provisional list of the job positions with the greatest risk of being replaced by softwares with LLM technology. The first fact that emerges from the research is that artificial intelligence concerns the whole spectrum of the job market. Researchers calculate that about 80% of the US workforce will see AI impacting at least 10% of the jobs performed. A percentage that for 16% of the workforce will rise above 50%. Overall, the paper suggests, 15% of all tasks currently performed by US workers could be replaced – with considerable time savings – by LLM applications.

At present, the study recalls, softwares based on generative artificial intelligence are capable of performing translation, classification, text, and computer code creation operations at excellent levels. It is therefore easy to say which are, to date, the job positions most at risk of being replaced by softwares: accountants (the most at risk according to the study), interpreters, mathematicians, and writers.

The research certainly does not have the ambition to predict how the labour market will evolve in the coming years, but it puts some cards on the table. These cards should receive a lot of attention from governments in the first place, but also from businesses and workers. An enormous effort is needed in terms of training, outplacement, and reorganisation of work. Rearguard battles will be of little use.

Featured image: retrieved from Adobe Stock

More from Author



Please enter your comment!
Please enter your name here