The race for artificial intelligence soars salaries and empties university chairs
It will not have escaped your notice that the giants of the tech have started a race for artificial intelligence: smartphones with facial recognition, voice assistants, autonomous vehicles, robotics, uses are not lacking … And to carry out these projects, these companies are using the means, in particular by offering salaries that reach unprecedented heights, even for an industry that has never been afraid of excessiveness in terms of the salaries of its talents. This is the subject of a recent survey conducted by the New-York Times.
AI experts paid like professional athletes
The typical specialist in artificial intelligence is usually a doctor fresh out of university, or if less qualified, with only a few years of experience. He is paid between 300,000 and 500,000 dollars a year or more, in salary and shares. For references in the sector, the package can be negotiated several million dollars, for contracts of four to five years, which they can then renew like a professional athlete. A relevant analogy, since some even say, half-jokingly, that caps should be established in the image of what exists for the players of the National Football League.
The most sought-after profiles are managers who already have experience in AI project management. So, during the ongoing trial against Uber, Google revealed that Anthony Levandowski, at the time one of the senior executives of Waymo, the autonomous car division of Alphabet, had received more than $ 120 million in compensation. , before joining Uber last year, through the acquisition of Otto, a startup he co-founded.
Less than 10,000 profiles in the world capable of performing level AI research
How to explain this inflation? First, the auto industry is in competition with the tech giants: the two find themselves coveting the same experts for the autonomous car. Companies like Facebook or Google are also considering AI as a way to solve various strategic challenges, from the development of digital assistants and smart home, the detection of offensive content. Most importantly, in the context of the current talent shortage, companies are trying to attract as many as possible. According to Element AI, an independent laboratory in Montreal, there are fewer than 10,000 people in the world capable of doing level research on the subject. According to Andrew Moore, dean of the School Computer of Science at Carnegie Mellon University: “It’s not necessarily healthy for society, but for companies, it makes perfect sense: they are eager to secure the talents to build the teams that can work on these technologies. “
In 2014, Google bought DeepLab, a 50-person artificial intelligence lab, for $ 650 million. In 2016, DeepLab had 400 employees, while, according to DeepLab’s financial report, the personnel costs were $ 138 million, or $ 345,000 per employee.
In artificial intelligence, the most advanced research concerns deep neural networks, which are able to learn tasks by themselves by analyzing data. The concept dates back to the 1950s, but until recently it was of little interest to companies and universities.
In 2013, Google and Facebook began to recruit the few researchers who had specialized in these subjects. It’s the neural networks that now help recognize faces in photos posted on Facebook, do speech recognition on Amazon Echo, or instantly translate foreign languages, like on Skype. Using the same mathematical techniques, researchers are improving autonomous driving, or developing scanners capable of detecting diseases, assistants capable not only of recognizing words but of understanding them, automated trading systems, robots capable of choosing products. without having seen them before …
Golden bridges for university professors
With so few specialists available, big tech companies are turning to universities to recruit. But in doing so, they cut into the number of teachers available for teaching. Uber hired 40 people from a recognized program at Carnegie Mellon University to work on its self-driving car project. In recent years, four of Stanford’s best-known researchers have left or taken leave of their chairs. In Washington, six of the 20 professors of artificial intelligence who are on leave, full or part time, and now work for private companies.
Some find a way to compromise. Luke Zettlemoyer of the University of Washington turned down a job in a Seattle lab run by Google that would have allowed him to triple his current salary (around $ 180,000). Instead, he chose a position at the Allen Institute that allows him to continue teaching.
Another way to deal with the shortage is training. Companies like Google and Facebook organize courses to teach their employees deep learning and related techniques. Fast.ai, a non-profit organization, or the Deeplearning.ai company, founded by a former Stanford professor who helped create the Google Brain lab, offer online classes.
The basic concepts of deep learning are not difficult to understand, requiring little more than a mastery of high school math. But real expertise requires a much higher level of mathematics and a form of intuition that some even call “Dark art”. And specific knowledge is also required for areas such as autonomous cars, robotics, healthcare, etc.
A talent shortage that is likely to last
Under these conditions, small businesses find it difficult to compete. In order to keep pace, they look for talent in unusual places. Some hire physicists and astronomers who have the necessary math skills. Other American startups are looking for employees in Asia, Eastern Europe and wherever wages are lower.
Except the industry giants are doing the same. Google, Facebook, Microsoft and others have opened labs in Toronto and Montreal, where much of the research outside of the United States is done. Google is also hiring in China.
Not surprisingly, many believe the talent shortage will last for several years. And according to a professor at the University of Montreal and eminent researcher, things are not likely to get better, for one simple reason: “It takes several years to train a doctor. “
Qu’est-ce que l’intelligence artificielle (IA)? Qu’est-ce que cela signifie pour les entreprises ? Et comment votre entreprise peut-elle en profiter ? Pour en savoir plus, inscrivez-vous au séminaire AI : Future of Business