Dec 5, 2018 - Technology

AI talent is not monopolized quite yet

Illustration: Rebecca Zisser/Axios

Early in the high-stakes race to dominate artificial intelligence, Big Tech — flush with cash, data, and name recognition — has seemed to have already captured AI's commanding heights and created an insurmountable, monopolistic advantage.

But new data suggests that the contest is not quite over, and that the field is much more crowded than was thought.

What's going on: Tech companies, startups, legacy companies and academics around the world are in a pitched battle to attract a relatively small number of talented AI experts, a struggle in which the chief weapons are money, prestige and glory.

  • Until now, this story has been dominated by dramatic incidents like Uber raiding the entire driverless research unit of Carnegie Mellon University, and Big Tech buying up startups by the dozen.
  • The resulting impression has been that AI talent is highly concentrated in Google, Amazon, Baidu and a few other Big Tech companies.
  • But a study by Diffbot, a Silicon Valley machine learning startup, has found that, even if Big Tech does employ a lot of AI experts, hundreds of thousands more are dispersed across companies the world over.

Realistically speaking, Big Tech simply cannot vacuum up the talent everywhere, says Diffbot CEO Mike Tung. "There are many places in the world where these companies simply have no offices or open jobs," Tung tells Axios.

Why it matters: That there is at least somewhat dispersed talent means that folks outside of Big Tech have at least a fighting chance to make the big breakthrough.

By the numbers:

  • More than 720,000 people worldwide have AI skills, by Diffbot’s count.
  • Just ten companies employ 10% of them.
  • But another 100 companies employ at least 1,000 people with AI skills. And more than 750 companies each employ at least 200 people with AI skills.

Even among the top rung, there are surprises, like Indian tech giants Infosys and Tata.

These numbers come from Diffbot’s whole-web search for every person with apparent AI skills, demonstrated through published academic papers, their LinkedIn profile, or their personal website.

Yoshua Bengio, a pioneering AI researcher at the University of Montreal, told me he's not surprised by the number of companies who have hired AI workers. Interest in AI "is springing from all quarters," he says.

  • Yes, but: Bengio cautions against reading too much into absolute numbers. If you take account of papers published for leading academic conferences, Big Tech again looks seriously formidable, he said.
  • "Anybody can say they have ML skills," said Oren Etzioni, CEO of the Allen Institute for Artificial Intelligence. Ranking companies by the quantity of AI practitioners on staff isn’t the same as asking which are best at it.
  • Tata doesn’t have a head-and-shoulders advantage over Apple in AI, Etzioni says, even if it employs 2.5 times more self-described AI experts, as Diffbot reports. And IBM’s expertise is "very limited," he says, despite its second-place ranking.
"There is a monopolistic tendency in the tech world, which might get worse with AI because of the winner-take-all advantage of having access to most data, talent, customers and cash (e.g. to buy competing start-ups). I'm not sure how to deal with all that but clearly this deserves a social and political discussion."
— Yoshua Bengio

Go deeper: Academia and the tech industry feud over AI talent

Go deeper