When the wave of generative AI hit, the wildest predictions arose about its consequences on employment. Hundreds of millions of positions were allegedly at risk. The dust is now settling: companies are learning to evaluate these tools, while consolidation begins among start-ups operating in this sector, as evidenced by the fire sale acquisition of Inflection AI by Microsoft. Logically, more reasonable analyzes are beginning to emerge.
The World Economic Forum estimated that generative AI would have a net positive impact on jobs leading to the disappearance of 85 million of them by 2025 but the creation of 97 million positions. These figures, over such a short interval, seem unrealistic for those who know the time of companies’ recruitment decisions, but the essential point is not there. The much-publicized quantitative analysis does not capture the power of the internal transformation brought by these technologies.
Because it is the way in which AI will recompose relationships within organizations, and de facto the organizations themselves, that is interesting. To begin with, these tools are very expensive, due to the computing power they consume. The price for Microsoft Copilot is 28 euros per month, compared to around 6 euros for the Office 365 subscription. A fierce debate is currently underway between the human resources departments and the financial departments of the CAC 40 to determine which grades employees must have access to it. This issue is exacerbated by the banning or non-support of these tools on a professional basis, while employees use them on a personal basis.
More autonomous employees thanks to AI
Let’s get over the price barrier, which should lower as open source models are optimized and disseminated. Generative artificial intelligence gives individuals autonomy over tasks previously reserved for central functions, such as the production of marketing content. It will likely increase the feeling of independence, particularly in flexible organizations that rely on outside workers, such as real estate agent networks. The temptation may be great for the latter to regain control of communication, for example by modifying the graphic charter or by erasing the headquarters logo from presentation documents.
People with advanced expertise – in sales, technology, etc. – but whose mastery of language and communication with others has until now been limited will also be able to fill these gaps thanks to artificial intelligence. In addition, the access to knowledge that AI allows reshuffles the cards for experts who were not directly involved in production tasks, but were paid to transmit specific knowledge. Why call on the little computer genius when I only have to ask the question in natural language to create a complex interconnection between Excel and another tool?
Access to data risks, finally, being disrupted. Because it is difficult to put in place fine-grained restrictions, the decision will become simple: either we block everything, or we give access to everything, by massively expanding the amount of information shared. Professionals whose strength lay in their knowledge of management tools to obtain this or that figure will thus see their usefulness contested.
This horizontalization will accelerate the questioning of hierarchy. With expanded access to data that was previously only communicated to them upon formal request, employees who had little say in the matter will be able to question the decisions made by their manager, or even make them themselves. The new extended language models (LLM) are multilingual and capable of understanding spoken language. No need to be a specialist prompt, a simple one-hour training course is enough to learn how to use these tools effectively. Bosses, be ready for the artificial intelligence revolution: it is not what you think.
*Robin Rivaton is Managing Director of Stonal and member of the Scientific Council of the Foundation for Political Innovation (Fondapol)
.