Jeff Holden, Chief Product Officer at Uber, made two announcements on the company’s blog the other day.
“The first is the creation of Uber AI Labs, a new division of Uber, based in San Francisco, dedicated to cutting-edge research in artificial intelligence and machine learning. And the second is the acquisition of the AI research startup Geometric Intelligence, whose 15 members will form the initial core of the AI Labs team.”
He continued later in the post: “The formation of Uber AI Labs, to be directed by Geometric’s Founding CEO Gary Marcus, represents Uber’s commitment to advancing the state of the art, driven by our vision that moving people and things in the physical world can be radically faster, safer and accessible to all.”
According to Bloomberg, GI had 15 employees. The company was creating cutting edge machine-learning methods. The new team will enhance the technology Uber utilizes for its driver – passenger pairings. Uber has an interest in self-driving cars as well. The amount paid for the startup was not disclosed. The New York Times, however, valued the deal near $70 billion.
Uber has brought in outsiders to buff its technology research efforts in the past. The company acquired two other companies recently to improve its in-house technology. The mapping tech startup, deCarta was its first acquisition. An Uber spokesperson commented on the acquisition to TechCrunch:
“A lot of the functionality that makes the Uber app so reliable, affordable and seamless is based on mapping technologies. With the acquisition of deCarta, we will continue to fine-tune our products and services that rely on maps – for example UberPOOL, the way we compute ETAs, and others – and make the Uber experience even better for our users.”
Its second acquisition was Otto; Otto created self-driving car technology.
Big Investments in AI for Uber
Dr. Gary Marcus understands Silicon Valley’s new focus on artificial intelligence. For example, AI has become an important frontier for companies like Facebook.
“Every major company realizes how essential A.I. is to what they’re doing,” Dr. Marcus said in an interview. “Because of the scale of data people are operating on, even the smallest gains in efficiency can turn out enormous changes at these companies, especially in terms of profit.”
In particular, companies pour money into deep learning techniques. These methods give computers tons of information to identify patterns and acknowledge and group objects.
In an interview with BBC, Dr. Marcus felt this approach was insufficient.
“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.” His method is the first step to make computers more human-like. GI employs several methods including Bayesian and so-called evolutionary methods. 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.” That’s why Marcus wants to get the success rate to 100% to prevent accidents or other self-driving mishaps.
For this reason, Jeff Holden thinks it will take some time before AI reaches the level to operate at the level of a typical Uber driver. But the solutions, he says, “are all going to come in the form of artificial intelligence.”