LaborIA: what's the latest on the artificial intelligence laboratory set up by the Ministry of Employment and Inria?

There are two main historical drivers behind the creation of LaborIA. The earliest corresponds to a recommendation in the Villani report of March 2018, which on the "work" section refers to the need to set up what it calls a "perennial research head", for both research and experimentation, to better understand the transformation of work by artificial intelligence in the face of a market that is not necessarily prepared. The second, in 2020, is the Global Partnership for AI, in which Inria is a major player, and for which France is leading, among other things, a working group dedicated to the future of work. This working group mentions the need to set up national Living Labs around the issue of work, in order to encourage experimentation and improve knowledge through practice.

In November 2021, Elisabeth Borne, then Minister for Labour, Employment and Integration, signed an agreement with Inria to create "LaborIA", a laboratory dedicated to artificial intelligence and its effects on work, employment, skills and social dialogue.

This laboratory, which is part of Inria's AI mission (led by Jean-Michel Lefèvre), is a place for research and reflection shared by players from all backgrounds, around one subject: work transformed by AI.

It's an imminently technical subject, but one with a social impact. Our aim is to understand how artificial intelligence can enhance or alter work as a means of acquiring material dignity, i.e. a decent income, and spiritual dignity, i.e. a meaningful activity.

We are doing this ambitious work in close collaboration with the Ministry of Labour. It's not research dedicated to feeding academic thinking, but in support of a public policy need. We set the research agenda and select the research areas together. We (the researchers) are the bearers of knowledge and the watchdog on the subject, so we are the driving force behind proposals on the issues, we provide the initial direction, and then we build on this with all the stakeholders.

The idea of including different stakeholders in the projects is to ensure that the issues identified are of real relevance to the people who work full time in this field, so as to produce really useful responses.

What we have identified for the roadmap for the coming years is that we no longer really have a sector where we can say that it has been transformed more than another by artificial intelligence. We realise that we are looking at a very broad spectrum. So we've decided to focus on two sectors: logistics and production, on the one hand, and the cultural and creative industries, on the other.

We're going to look at them independently over the next three years, but we're also going to mirror them, to see the similarities and differences between the two sectors. Our first focus will be on 'skills and training' and 'working conditions'.

What sets us apart from the major OECD studies is that we work very much around this notion of 'real' work. In other words, the major studies estimate the number of jobs exposed and sometimes destroyed by constructing an exposure index based on the state of the art of technology compared with the nomenclature of skills for a particular occupation. It's all quite theoretical, in fact.

Our second area of work will be to look at cross-functional issues. For example, we're going to look at AI integrated into recruitment processes and their impact on the recruitment process, with a particular focus on the potential known biases of artificial intelligence. The aim of this area is to be able to respond to requests from public or private organisations to address their AI issues. This will enable us to test the operationality of the results we have developed, and in return the situations in the field will help us to better design our research outputs.

Finally, the third axis of our roadmap will focus on the creation and highlighting of tools (methodological sheets, thematic sheets, workshop proposals, video capsules, etc.), based on the studies carried out during the first two years of LaborIA's life. The idea is to give everyone the opportunity to benefit from useful resources.

The first two years of LaborIA's launch, called Explorer and operated by Matrice, were devoted to research and experimentation.

We published an initial questionnaire survey of 250 SME managers, whom we questioned about the quality of life at work with AI. Of these 250 managers, 50 responded on the basis of experience already gained with AI, and the other 200 on the basis of their own representation of what quality of life could be with AI, but without experience. This enabled us to highlight one thing: when you've already had experience of artificial intelligence, you ask yourself very different questions than when you've never had any. In the latter case, the people interviewed are generally wondering about dehumanised work, while others who have already used it are wondering about autonomy and know-how.

When you've tried it, you're more factual, whereas when you haven't, you're more emotional. Fantasies are not adjusted to what is really happening.

We also carried out two levels of study around the notion of use cases. We identified a variety of use cases, such as load calculation systems or aircraft engine maintenance systems, and then studied them by interviewing users to get a snapshot of the social situation at a given moment in time. Another approach was to study them over a year. This enabled us to see the evolution of user behaviour in relation to AI, from refusal at first to acceptance at a later stage.

Over the last two years, we have also set up on-boarding seminars aimed at specific communities to help them mature their AI projects, and in particular the social dimension of AI.

Finally, we have set up a Comex extended to include the social partners, to listen to their views on the subject and present them with the results of our studies. The idea is to create a real dialogue around this technological transformation.