Private industry raids universities for artificial intelligence engineers
SEATTLE -- The shortage of experienced artificial intelligence engineers is so severe that top tech firms are luring professors out of academia with seven-figure salaries.
It’s posing a dilemma for the industry: Who will teach a new generation of AI students the tech industry desperately needs?
Professors who have spent much of their adult careers researching AI -- and some never having worked in the private sector -- are being pursued with offers of million dollar stock incentives, cash and preferable working conditions.
“As a result of that we started actually seeing major players in the industry start making insane offers,” says Ali Farhadi, an associate professor of computer science and engineering at the University of Washington.
“These days seven digits is not out of the norm,” says Farhadi, who admits to receiving million dollar offers.
Luke Zettlemoyer has also seen big offers.
“I think you can go to three or four times your university salary often pretty easily,” says Zettlemoyer. who is also an associate professor of computer science and engineering at the University of Washington.
State salary records show some UW professors with an expertise in AI are making upwards to $180,000 per year. But with that salary comes years of research work at some of the most prestigious universities in the world.
Zettlemoyer received his Ph.D. from MIT and did post-doctoral work at the University of Edinburgh.
Hannaneh Hajishirzi is another professor who has seen big offers to leave academia. The UW research assistant professor of electrical engineering believes the demand for professors could hurt the industry going forward.
But she, along with Zettlemoyer and Farhadi, share a big common theme. All of them have turned down the big -money offers.
Hard numbers about the shortage of experienced AI engineers are difficult to find, but professors are seeing more and more temptation to join the private work force.
“Basically companies are reinventing themselves as artificial-intelligence or machine-learning companies because there is such a high demand,” says Zettlemoyer.
Which is posing the dilemma for academia and industry. If companies lure the experienced teachers away, students could suffer.
“There is just more and more of a need, the university programs can’t grow fast enough to keep up with it,” says Zettlemoyer. “If you take everybody ( teachers ) all at once, of course, everything would collapse and can we sustain the pace that's going on right now? We'll see, but it is definitely an issue.”
“That’s an important fear,” says Hajishirzi.
As an alternative, several UW professors have been able to keep one foot in academia and one foot in the private sector.
Both Farhadi and Zettlemoyer work part-time at the Paul G. Allen Institute for Artificial Intelligence. It’s a Seattle-based nonprofit research firm of AI professionals who work on industry projects, but release their work publicly, free of charge.
They are just missing the million dollar salaries.
“I love this arrangement,” says Farhadi. “It’s the best of both worlds. On the Allen Institute side, I get the resources I want to launch big projects, and on the university of Washington side I get the freedom to do things.”
Farhadi find personal satisfaction in teaching, something he believes he would miss if he went to work full time Google, Amazon, Microsoft who are fighting for experience AI talent.
“To some of us that’s a little too restrictive meaning that you may not want to commit to a specific product,” says Farhadi.
“There’s very much a danger of killing the goose that lays the golden egg,” says Oren Etzioni who left his professorship at the University of Washington in 2014 to become CEO of the Allen Institute for Artificial Intelligence.
“Our university system is at the core of training the people,” says Etzioni. “Now if you hire away the teachers and you hire away the students before they even finish their training, that's a problem.”
Zettlemoyer feels the same way and chose personal satisfaction of teaching the next generation artificial intelligence while working on projects he finds beneficial for the entire industry.
“The whole point of research in this area is to make computers do things they couldn't do six months or a year before,” says Zettlemoyer. “That’s kind of amazing, and it’s not clear how far can we push that, but to me that's more valuable than money.”
Other professors have found similar arrangements with large tech firms. Hajishirzi consults for several firms and continues to teach.
“We try to play nice with folks like professors at the University of Washington and not steal them” says Etzioni.
Who would have thought a job making computers think like humans would be on par with the salaries of professional athletes?
There’s nothing artificial about that.