At a time when AI automates knowledge, mathematical rigor is becoming more strategic than ever.
At a time of development of generative artificial intelligence, it is urgent to put mathematics back at the center of learning. An apparent paradox emerges: while ChatGPT can instantly give the date of birth of Louis XVI, the composition of atoms or the GDP of Brazil, factual knowledge loses value. However, this technological revolution does not signal the end of intellectual skills. On the contrary, it redefines its hierarchy.
Basic knowledge devalued, expertise valued
The 2019 baccalaureate reform caused an unprecedented collapse in the number of students taking mathematics at general high schools in France. Scientific numbers have fallen by 30% for boys and 60% for girls, bringing the proportion of science baccalaureate graduates back to the 1965 level. A historic decline that comes at the worst possible time.
Because if basic knowledge becomes commoditized by AI, two profiles emerge strongly. On the one hand, experts capable of giving precise and structured instructions to artificial intelligence. On the other hand, professionals able to challenge and validate the responses produced by these systems. Between the two, a worrying gray area where basic factual knowledge is no longer enough to create value. The era of instant responses now demands analytical skills that only rigorous training can develop.
The hidden equation of effective prompting
The daily use of generative AI reveals a disturbing truth: the quality of the prompt determines the quality of the response. However, building a good prompt does not come down to mechanically following four or five methodological steps. This requires a spirit of analysis, synthesis and structure. The more a user masters the art of breaking down a problem, identifying relevant variables, and organizing their thoughts, the more capable AI becomes in their hands.
These cognitive abilities do not fall from the sky. They are mainly formed by learning mathematics, which trains the brain in logical rigor, modeling and structured reasoning. A student accustomed to solving complex equations naturally develops a methodology transferable to formulating sophisticated queries to an AI. Conversely, anyone who has never been confronted with the need to methodically break down a problem will remain a superficial user, incapable of fully exploiting the potential of these tools.
France in reverse
Three years after the reform, the number of hours of mathematics taught by high school teachers fell by almost 20%. The removal of mathematics from the common core of general first and final year left 150,000 students without mathematical training in final year in 2020-2021, compared to 40,000 before the reform. The pool of mathematical talent is drying up at precisely the moment when global technological competition is intensifying.
The consequences go far beyond the school environment. In a context where the number of job offers requiring AI skills has increased sevenfold between 2018 and 2023 in France, training a generation of students who are deprived of mathematics amounts to jeopardizing our technological sovereignty.
The race for technical talent is accelerating
Beyond the daily use of generative AI, a global race for technical talent is raging. Creating, training, and optimizing AI models requires advanced mathematical skills in linear algebra, probability, and statistics. But demand is no longer limited to engineers and researchers alone.
Tomorrow, lawyers will have to understand the legal implications of algorithmic decisions. Managers will need to master data analysis to manage their teams. Professionals in all sectors will need to collaborate with AI systems whose operation relies on mathematical concepts. This profound transformation of the labor market requires a solid foundation in mathematics for the entire working population, not just for a technical elite.
Urgency of an educational burst
The French situation should raise alarm. Not only are we training fewer mathematicians, but we are also increasing inequalities. In CM1, 15% of French students do not master basic mathematical skills, compared to 7% on average in the European Union. Only 3% of French students are considered high achievers in mathematics, compared to 46% in Singapore.
Students who do not do mathematics will develop significantly weaker analysis, synthesis and calculation skills. They will quickly be overtaken by those who have mastered both AI tools and the mathematical thought structure needed to fully exploit them. In an economy where competitive advantage is based on the ability to increase one’s intelligence through technology, the absence of mathematical training condemns one to marginalization.
It is time for a radical change of strategy. Reintegrate mathematics into the common core, revalorize the mathematics teaching profession, modernize programs to make them more applied to AI issues: the projects are immense. But the alternative is simple: train generations of passive users of tools they do not understand, or give future professionals the keys to mastering and shaping the era of artificial intelligence. The choice should be obvious.
Martin Pavanello
