"It's urgent to foster acclimation so we don't miss this train"
Mustapha Benkalfate first introduces himself through his area of expertise: the acclimation of non-technical professionals. At the time of the interview, over 50,000 people had learned to use AI with his methods via LinkedIn Learning and Microsoft.
"My expertise and primary focus are on AI acclimation and its mastery by non-technical professionals."
The under-discussed advancement: what AI does to our way of thinking
When asked about the innovations that have marked him, he dismisses the noise of announcements to highlight a then-under-discussed topic: studies measuring the effect of AI interaction on cognition.
"One of the key aspects that isn't discussed enough is recent studies revealing how the cognitive interaction between humans and AI profoundly transforms our ways of thinking."
This theme has become central to his work. He has since developed it in his analysis of studies on cognitive debt and in the eight cognitive traps of generative AI.
Employment in France: skepticism prevents proper adoption
Regarding the impact of AI on employment in France, his response is more a critique of the public debate than a prediction. According to him, the national penchant for abstract discussion delays practical learning, without preventing its use from spreading.
"Skepticism doesn't prevent adoption; it prevents proper adoption."
"I believe we must train people thoroughly, always inform them of the risks, but then, we must rely on their intelligence rather than scaring them."
He already described the problem he is tackling in companies today: the use of AI tools outside the rules set by the employer, involving sensitive data. "It is already a documented reality in France," he says, and "a challenge to stop these already massive practices." This is shadow AI.
Ethics: The Illusion of Mastery
The ethical risk he highlights is not a spectacular one. It is the impression of competence that the tool gives after just a few exchanges.
"Generative AI gives the impression that you've mastered it after a few exchanges. This is false and dangerous."
In July 2026, speaking to faculty at the University of Nottingham, he gave a name to this phenomenon among leaders: the "Dunning profile," stuck at the stage of unconscious incompetence because ChatGPT answers everything.
Eighteen months later, a confirmed stance
Since the interview, two facts have lent weight to his positions. The MIT NANDA report established in 2025 that 95% of organizations derived no measurable return from their generative AI projects, due to a lack of integration into actual work (our analysis). And the AI literacy requirement of the European AI Act has applied since February 2, 2025, to companies that deploy AI systems. Mustapha Benkalfate co-founded and now leads Revenu IA to address these two findings.