Evangelis-IA-tion: giving tools to people so they don’t become machines

Generative Artificial Intelligence (GenAI) has given rise to fears that human beings are losing ground to technology. To allay these fears, companies are getting their act together. In a more or less formal way, they are educating their employees about AI. A way of (re)placing people at the heart of the GenAI.

By Kevin Erkeletyan

“We choose France.” With these words trumpeted one by one, on 13 May 2024 Microsoft Chairman Brad Smith announced a €4 billion investment in France. Actions include a plan to train one million French people in Artificial Intelligence (AI) and to support 2,500 local start-ups in their adoption of AI by the end of 2027.

This is not overambitious, according to Pascal Bizzari, Deputy Managing Director of AVISIA, a consultancy firm “committed to the AI revolution in business.” What immediately struck him when generative AI came on the scene was “the lack of support in organisations.” In some, “it was the Wild West” with a laissez-faire, laissez-passer attitude; in others, “all AI tools were held at bay.” Since the emergence of GenAI, his firm has been regularly called upon to “demystify” the issue among other things. This “change management” begins by dispelling misconceptions as a priority.

Eneric Lopez, Director of the National Initiative on AI and its social impact at Microsoft, France “Anyone can use a screwdriver and a size 12 spanner, but it doesn’t mean they’re cabinetmakers.” In short, AI is like anything else: it has to be learned.

THE STONE IN THE SHOE THEOREM

In companies, the AI wave sometimes comes up against the odd breakwater. Eneric Lopez cites the example of his finance department. When he was wondering about rolling out machine learning in certain processes, he was confronted with two reactions: “a machine is not going to teach me my job” and ”management wants to replace us.” “Which just goes to show that we’re like any other company.” For him, this need for training has to answer one question: “Why do it?” A question to which he has a simple answer: ‘What’s the stone in your shoe? And this is perhaps what differentiates the AI-tool from the AI-machine: function. So change, he says, has to involve two phases: awareness-raising and desire. Making people understand and getting them on board; making people understand and thus making them want to do it.

Companies take different routes to achieve this. One digital player has set up its own in-house school: all employees can receive training in AI according to their level and needs. The basic idea is “active adoption”: making the toolbox available to everyone and letting each person decide whether or not to use it, at the risk of widening the gap between the real techies and those who are less so. At Roquette, the aim is to train all employees. “Learning Thursdays” invite them to work on AI and, “every year”, Pierre-Louis Bescond’s Data team “creates over 1,000 hours of training in the form of quizzes, games and interactive videos.” He has also trained himself in Artificial Intelligence, following the Stanford MOOC “in the evenings and at weekends.” And more broadly, he is a member of the governance committee of the Cité de l’AI, a Medef Lille Métropole initiative involving local businesses.

When you no longer understand what’s going on in the machine, then you’re being subjected to something; you’re in a situation of alienation, and the production process is out of your hands. And that’s where AI shifts from tool to machine.

Luca Paltrinieri, Researcher in Philosophy

Everyone is talking about “acculturation”. And everywhere, it goes down less well-trodden paths. Jean-Luc Leblond and Jean-Baptiste Girardin, Data Product Managers at Malakoff Humanis, work in close contact with the group’s various divisions. They “evangelise” every day. It’s not their core profession, but it’s also what it has become. Jean-Baptiste Girardin talks about a “cultural revolution”. Explaining, organising meetings and “helping people to understand the ins and outs of AI” is all about probability, not certainty. “At one point,”, he says, “the main part of my job was to explain that we were going to incorporate randomness into processes.”

Béatrice Matlega is addressing an entirely different target: an audience that asks “why?” “Why isn’t the link responding? Why didn’t it say the same thing to me that it said to him?” The Director of Partnerships and Skills Programmes in Education at Microsoft “trains the younger generation and future talents in algorithmic thinking.” And she has started with some 10 to 11 year-old pupils at a school in Issy-les-Moulineaux, with whom she set up an experiment. “We got them to use Copilot to create visuals and a short text they could then present in front of a camera in a workshop, as an exercise in public speaking. The idea was not showing them how to use the tool, but getting them understand how to question an artificial intelligence system and develop their critical faculties as regards the content produced.” In short, to develop digital literacy. “At one time, this meant being able to use a computer, then to use office automation tools, then to have a basic understanding of code, and so on,” says her colleague, Eneric Lopez.

“It’s essential to develop a truly democratic digital literacy,” says philosopher Luca Paltrinieri. When you no longer understand what’s going on in the machine, then you’re being subjected to something; you’re in a situation of alienation, and the production process is out of your hands. And that’s where AI shifts from tool to machine. It’s no longer something that helps us, but something that dominates us.” The professor at the University of Rennes goes even further: to train or not to train in AI, and the way they do it, is a political choice for organisations. “Any technological tool can become a means of domination. Technology is the Pharmakon; it’s both cure and poison. It all depends on how it’s applied.”