Ride-sharing app acquires start-up Geometric Intelligence

Ride-sharing app company Uber may be looking to accelerate self-driving car development with the acquisition of Geometric Intelligence, a New York-based start-up.

The acquisition is expected to see Uber gain 15 specialist researchers who will form the backbone of the new Uber AI labs – a new division formed by the ride-hailing service and set to be work on a range of projects from self-driving software to predictive analytics.

Speaking to the BBC, Geometric Intelligence’s co-founder Gary Marcus said that the Uber AI Labs team will have a wide brief, from flying cars to improving traffic predictions.

However, Uber is also expected to put a renewed focus on the impact of automation and new technology on job losses – an issue occasionally raised with the company credited with inspiring ‘uberisation’ term and the disruption trend. Uber may be taking this approach as the rise of automation and self-driving cars continues to gather pace, with manufacturers such as Volvo already rolling out cars with self-driving capabilities.

Stating that it is less job losses and more shifts in workforce when it comes to technology, Mr Marcus said:

“Historically what you’ve seen is as technology has taken jobs away from people, new jobs have opened up.”

“The way I think about it is there are trade-offs, but one of the plusses is that eventually the roads are going to be much safer for people.”

Mr Marcus also told the BBC about the possibility of flying cars, which complements a recent whitepaper on the subject published by Uber.

“They’re going to allow people to take long commutes at 75 or 150 mph where you’d otherwise get stuck in traffic. It’s really, in a not-too-distant future, going to be something that is practical.”

Geometric Intelligence, which was acquired by Uber for an undisclosed sum, draws inspiration from how a child’s brain develops, an alternative AI approach which sits in contrast to deep learning investment made by the likes of Google and Facebook.

“Deep learning is not all that it’s cracked up to be,” Mr Marcus told the BBC.

“It’s very good for certain problems, but it doesn’t allow us to do the kind of inferences that people often do. We need next-generation techniques.”

“What I see is approximation with statistics. What people do is get something that works 80% of the time and they’re happy. Nobody dies if I tell you you’d like this book and you don’t.”

“When you work on a problem with a real-world consequence like driving, you want to be as close to 100% as possible.”