Uber creates AI Labs division to boost driverless research
Uber is pushing forward with its self-driving technology research, acquiring New York-based artificial intelligence startup Geometric Intelligence.
The ride-sharing service will use the company’s 15 specialist researchers to form a new Uber AI Labs division, aimed at improving traffic predictions and, according to the BBC, working towards a future of flying cars.
Initially, the new division will reportedly be tasked with improving the company’s ability to predict when and where cars are needed, but it will also be looking at ways of speeding up the launch of Uber’s self-driving cars.
Related: CES 2017
The company has been testing autonomous vehicles in Pittsburgh, where its Ford Fusion hybrid was spotted on roads earlier this year, and San Francisco.
Uber officially announced its self-driving tests in a blog post in May, where it explained the Ford Fusion cars would be “collecting mapping data as well as testing its self-driving capabilities”.
The firm is also testing Volvo XC90 SUVs that have been modified to run autonomously and have been ferrying around customers in Pittsburgh completely free of charge, as a thanks for testing out the future of road transport.
Thus far, the autonomous cars have been driven with a human behind the wheel in case anything untoward should occur, but Uber will want to launch fully autonomous technology sooner rather than later.
The company has reportedly been subsidising fares in recent years, which led to a loss of more than $2bn (£1.6bn) last year.
Related: Is the Tesla Model S the ultimate rental car?
Of course, the advent of driverless cars will mean drivers, and their salaries, are no longer required, with Uber’s chief product officer Jeff Holden telling the BBC: “Historically what you’ve seen is as technology has taken jobs away from people, new jobs have opened up.”
For now, Geometric Intelligence’s staff will move to Uber’s headquarters in San Francisco where they will continue research into a type of artificial intelligence that the startup’s co-founder Gary Marcus said goes beyond ‘deep learning’ – the technique being used by other companies, such as Google, to develop AI.
Instead, the startup has been using an alternative technique that will, according to Marcus, produce more accurate results when it comes to statistics than other approaches.
As he explained to the BBC: “When you work on a problem with a real-world consequence like driving, you want to be as close to 100% as possible.”
WATCH: Tesla Autopilot Safe? Head-to-Head
Let us know what you think of Uber’s latest move in the comments.