Talent shortage in AI is a myth: reply to Cedric Villani
Artificial intelligence recently made the headlines in France: the government is making plans in AI, with an initiative led by Cédric Villani, member of parliament and professor of mathematics. In his report, he proposes to develop education in AI, and to multiply the number of students by 3. He says there is a talent shortage.
I am surprised, because I am running 2 groups about jobs in artificial intelligence, one on Facebook (+2600 people) and another on LinkedIn (+500 people). On the other hand, there is only one job posting every month.
So after all, maybe there is a shortage of talented employers and managers. Instead of letting himself distracted by questions of education, Professor Villani should focus his energy on improving job market efficiency, and business opportunities.
There is no shortage of high-quality education in AI, especially online, and at all levels. For undergrads and below, there are Khan Academy, Coursera, EdX, Udacity, Kaggle, RAMP, Siraj Raval… for PhD-level and research, there are OpenAI, AI-ON, Ruder, and I am even running my own platform, StartCrowd.
I don’t even include the general-purpose Arxiv platform, which is sufficient for the most dedicated students.
As a result, I don’t think that the mission of the French state is to compete with online education platforms. It’s a lost battle, and a waste of taxpayer money.
Moreover, Professor Villani wants to make academic careers more attractive, by raising salaries. He’s worried about young talents moving to the USA. It’s a surprising idea coming from a seasoned academic like Villani. He is supposed to know that the academic job market is already very competitive, in France and elsewhere. For example, this year in France, there are only 17 job openings for junior full-time researchers (CR2 CNRS) in mathematics, Villani’s specialty, for hundreds of PhD applicants. If earning x1.4 the minimal wage was too low, there would be much less candidates. But applicants know that the CNRS in France is more attractive than adjunct faculty in the USA. Later in their career, competition is also tough: only 6 positions for senior full-time researchers (DR2 CNRS) in mathematics this year, corresponding to a junior/senior ratio of almost 3. That’s the equations that Villani should solve.
There is a shortage of scientists only on TV shows, a topic Villani can talk about better than me. In the real world, the global academic job market is flooded with PhDs, since the eighties. It’s a structural feature of this market, akin to a Ponzi scheme.
In this context, if Google can relieve some of the pressure affecting the French academic system, Villani should be glad. However, Google has a very specific approach to research, which is not directly competing with academia. I discussed Google’s approach on my blog last year:
At the end, this Villani initiative has the wrong focus: too much science, not enough business (1.5 Billion Euros investments are peanuts). It’s relatively easy to get skills in AI, it’s much harder to make money out of them. That’s why selling courses is one of the most effective businesses in AI now, as acutely observed by the French data scientist François Chollet (disclosure: Chollet is running the AI-ON education platform, and I am running a competitor Startcrowd):
Update: interesting reactions on Reddit Math
