From University to Industry: The War for AI Talent

By Matthew Peroni

In September, IBM announced a $240 million joint Artificial Intelligence (AI) research effort with the Massachusetts Institute of Technology. The investment follows a recent history of underwhelming performance from IBM’s Watson health program, which made big promises of using AI to tackle humanity’s most dangerous and elusive diseases. Spending on marketing for Watson has been plentiful, but with few deliverables, the tech industry has been quick to dismiss the entire program as a disappointment. IBM, now aware of its deficiencies in AI talent, has tried to catch up with other tech giants, such as Google and Amazon, and has struggled mightily. What happened to all the AI talent?

With most of the successful AI startup companies already eaten up by Apple, Google, Microsoft, and Amazon, these major AI employers have decided to go back to school, raiding AI research labs on university campuses around the country and taking all of the promising talent they can find. In January of 2017, Google acquired Dr. Fei-Fei Li, who was working at Stanford University as a professor and director of the Stanford AI Lab. At Stanford, Dr. Li created the Visual Genome, a database of meticulously labeled images used for training AI systems that is free and open to the public.  In September of 2016, General Electric purchased a small Berkeley-based machine learning startup called which, in very Berkeley fashion, sought to “democratize AI.” Even further back, in January of 2015, Uber acquired entire research groups from Carnegie Mellon’s National Robotics Engineering Center. In an interview, a previous employee at the robotics center expressed serious concern, asserting that “there are absolutely no checks on the work that they are doing or what they are taking. Is it for CMU? Is it for Uber? None of us here know.” Even for investors in one of these tech companies, this should be worrying.

For the companies that were ahead of the curve and acquired this AI talent early, such as Uber and Google, this is all a good thing. They now own scarce expertise and will likely see large returns on these investments. For everyone else, the question remains, what could have been? What would the startups have produced following their own missions? What would the research groups have created and made available to the public? AI’s future remains unclear, but now that companies are going after university research, it is no longer just a matter of which company will get the biggest piece of the pie. It is a matter of who owns the future of computing.