Quantcast

Uber’s AI Hub in Pittsburgh Gutted a University Lab—Now It’s in Toronto

Jordan Pearson

Jordan Pearson

Will Uber be a partner in Toronto?

On Monday, Uber announced that the company is opening an AI research hub in Toronto for self-driving cars—its first outside of the US.

The news was greeted with celebration, since wooing Uber is a good sign that investments in the newly-opened Vector Institute for AI research are already paying off. Uber is also investing $5 million in Vector, which is affiliated with the University of Toronto, putting it in league with the likes of Google and Canadian banks, who are also sponsors. But for those who remember what happened when Uber opened a branch of its Advanced Technologies Group (ATG) in Pittsburgh, the move is likely to be welcomed with caution.

In 2015, Uber announced a partnership with Carnegie Mellon University, which at the time hosted one of the top AI labs in the US. The company set up an ATG branch in Pittsburgh, near the school. Almost immediately, roughly 40 of the CMU lab's 100 researchers, including its director, went over to Uber. The lab's budget was slashed, and entire projects folded. As thanks, Uber donated $5 million to the school. No joint projects between Uber and CMU were ever launched, and the AI lab had to rebuild its bench strength with new hires.

Read More: The Head of CMU's Robotics Lab Says Self-Driving Cars Are 'Not Even Close'

Now, critics are wondering whether Uber will be a partner in Toronto, or a parasite draining brainpower (and taxpayer-funded research) from public institutions. The AI technology that Uber and other US companies use was born in Toronto, largely thanks to work by computer scientist Geoffrey Hinton, and the Vector Institute was set up to repatriate that knowledge. Competing with deep-pocketed corporations in a talent war is a challenge for schools, whether in Toronto or Silicon Valley.

"To have 100 people [at Carnegie Mellon] and to take 40 is rather dramatic," said University of Washington law professor Ryan Calo, who co-authored a paper on the pitfalls of the sharing economy, over the phone. "I do think that in many contexts Uber has been somewhat predatory, so it doesn't surprise me. Uber seems to live larger than life in this way."

But, Calo added, it's not a bad thing in and of itself for academics to move into the private sector. It's a question of balance.

"It sounds like they're providing a potential boon for local talent to stay in Canada," Alex Rosenblat, a researcher from the from the Data & Society Research Institute who co-authored the paper with Calo, told me over the phone. "The harm is: if you're going to rely on publicly-funded research from universities to fuel your proprietary software initiatives, there's no clear public benefit."

One of the ways that Uber can give back to the community in Toronto, according to Calo, is by sharing data that can help governments plan cities.

Read More: Uber Is Working to End Ridesharing

"Uber has access to [information on] where people are coming from and going to, potentially where traffic problems are in the city, [and] they could have access to things like speeding and parking information," Calo said over the phone.

"Google, Uber, and Facebook all benefit from public research that's done at schools, and they benefit from getting access to cities—is there reciprocity?" Calo continued.

A City of Toronto spokesperson confirmed to Motherboard that there is no agreement between the City and Uber regarding the ATG branch, and Toronto Municipal Licensing & Standards was not aware of the project.

As for how Uber will hire talent here, details are vague. The company has stated it will hire "dozens" of researchers. University of Toronto computer scientist and Vector Institute co-founder Raquel Urtasun will be leading Uber's hub and spending the large majority of her time there, according to an Uber spokesperson. Students will be both fulltime and part-time at the company, and students will be completing their PhDs while at Uber.

A Vector Institute spokesperson confirmed that there are no hiring or information-sharing agreements between the institute and Uber. However, Uber's sponsorship affords the company perks like attendance at Vector's summer training programs.

Calo noted that commercialization is a key part of bringing any new technology to the masses. That means private companies have to get involved. But, as Uber's actions in Pittsburgh showed, that relationship isn't always an equal one.

Subscribe to Science Solved It , Motherboard's new show about the greatest mysteries that were solved by science.