Careers under algorithms… will the future of work be fair ?

Automatically generated translation

Artificial intelligence is rapidly being deployed across our working environments, raising existential questions about the future of work. Managers now oversee hybrid teams made up of humans and AI agents, while a growing number of employees are questioning their added value. At the same time, recruitment managers are struggling to maintain the line between assisted decision-making and delegated decision-making. 

On 26 May, the Maison des ESSEC hosted a Careers Life Long Learning event on the theme “Careers under algorithms... will the future of work be fair?”. For the occasion, three experts came to share their analysis: Rébecca Renverseau, an engineer specialising in IT and innovation, Co-founder and COO of Tomorrow Theory, Jessica Her (E08), member of the IBM France Comex, head of the Talent practice at IBM Consulting and France lead for the IBM Culture programme, and Fabrice Marque (E95) Senior Strategic Advisor, business angel and founder of the ESSEC–Accenture Strategic Business Analytics Chair. 

In recruitment, AI is already at work. While large groups insist that the final decision still belongs to humans, what about the intermediate decisions? “Take the example of a recruiter who receives 800 CVs for a given position. The AI filters them on various criteria and brings five back to them. The recruiter shortlists three. Are they responsible for the three CVs they chose, or are they responsible for the other 795 that were rejected?”, asks Jessica Her. 

In practice, AI is merely a mirror, a reflection of what already exists. It therefore applies the same biases as those at work in society. The risk is one of a permanent reproduction of the past. “The AI will base itself on the historical data. So, in a company where white men in their thirties hold the most senior positions, the AI will replicate the same pattern and promote the same type of people”, explains Rébecca Renverseau. The biases can, however, vary from one model to another. “Have two recruitment profiles sorted by two different LLMs. You will get different results, which are a function of the way the AI was built”, points out Fabrice Marque. 

In the longer term, one of the pitfalls would be to fall into a “battle of machines”, warns Jessica Her, a recruitment world “where the dialogue would take place exclusively between machines, with AI-built CVs on one side and an AI agent tasked with sorting them on the other”. Under the AI Act (the European law on artificial intelligence), recruitment is now classified as level 3, that is to say “at risk”. It is therefore subject to oversight. Every filter applied must be justified by human choices that are demonstrated and owned.

A threat to women’s employment?

The experts agree that there are today two types of jobs particularly threatened by AI. On the one hand, transactional jobs (data entry, reporting), and on the other, those that require capacities for synthesis, analysis, comparison or in-depth research. In detail, women working in the services sector are the most exposed. “A study carried out by the International Labour Organization finds that jobs held by women in high-income countries such as France are three times more exposed than jobs held by men”, Rébecca Renverseau points out. For, in practice, the transactional, the routine and the quantifiable are all tasks carried out by women. “Historically, automation first concerned manual jobs, held by men. Today, AI changes the game: it is the skilled or skillable jobs that are at risk and weakened”, Rébecca Renverseau continues. In everyday working life, a divide is emerging between the employees most familiar with AI and the others. “You have to bear in mind that an expert learns much faster than a novice. In France, only a quarter of women hold a job linked to AI. So, if AI projects are always entrusted to the same people, a skills divide will set in and the gaps will widen”, warns Jessica Her. 

Within organisations, a paradigm shift is taking place. “What used to be expected in three days is now expected in half a day. The value of work is changing”, observes Jessica Her. In this context, teams may feel under pressure, while managerial skills atrophy. The risk, in the longer term, is not only that of a divide between those who master AI and the others, but of a silent erosion of know-how among those who delegate decision-making to it. “The gaps are not necessarily about who is the most productive, they can also be about who loses skills the fastest”, qualifies Rébecca Renverseau.