Uber has become infamous in recent months for all the wrong reasons, so in retaliation for being told how to treat people with a basic level of compassion, it has decided to accelerate autonomous vehicles.
In an interview with Bloomberg at Davos, Uber CEO Dara Khosrowshahi has made two bold claims. Firstly, the company will be profitable within three years, and secondly, autonomous vehicles will be part of the fleet within 18 months. It is a big statement to say Uber will have commercially active autonomous vehicles on the road in that time, but a lack of details will certainly help the CEO here.
Firstly, lets deal with the profitability one. Uber claims it is trimming the losses it is making each month, but many news outlets reported the latest quarterly earnings were north of $1 billion. CEOs have to be ambitious, but trying to reverse a loss of billions of dollars annually to a profit within three years is a big ask.
Of course, should Khosrowshahi be able to do away with those pesky employees, it would be a lot easier. This leads us very pleasantly onto the second claim made by the blue-sky thinking Khosrowshahi. Autonomous vehicles within 18 months.
This is of course a heavily hedged claim which was light on detail. Under ‘perfect’ conditions, an autonomous vehicle could operate in the near future. These ‘perfect’ conditions would be where the route has been perfectly mapped, while also taking into account other factors such as weather. In places like Phoenix, where Uber has been testing autonomous vehicles, this could be 5% of journeys within the 18 month time frame.
Of course, autonomous vehicles are all about the data which is collected in the city itself. Once you have mapped the city perfectly the first hurdle has been conquered. Considering the number of vehicles it has in the major cities of the world, there are few companies who are in a better position to accurately map the environment. Monitoring these vehicles will also collect information on how an autonomous driving programme should react to certain situations.
In theory, Khosrowshahi is right. His worldwide fleet could be collected all sorts of information to train AI to drive, but we are highly sceptical about whether there will be autonomous Uber vehicles operational in 18 months, even in rare circumstances, or the five year target Khosrowshahi seems to have set for wider usecases. From a technology perspective, we might be ready, but the other cogs are unlikely to have been aligned.
The insurance industry will have to have a major reform, cellular connectivity will have almost ubiquitous coverage, laws will have to be changed, councils will have to get better at planning and communicating work, consumer attitudes will have to change and there are countless other parallel challenges which are out of the reach of software engineers.
We can imagine autonomous vehicles will be on motorways or major highways sooner than some think, but these usecases will only be ones where pedestrians aren’t. As soon as there is an opportunity for unpredictable behaviour, regulatory restrictions will take years to change. Ultimately, autonomous vehicles will not be present in the cities until this question can be answered:
An autonomous vehicle is driving down the street when a six year old child runs into the path of the vehicle. The vehicle is moving at such a speed that stopping prior to a collision is not a possibility. Does the vehicle:
- Swerve right into a wall and injure, potentially fatally, the person in the vehicle
- Swerve left and hit, potentially fatally, an elderly couple walking along the pavement
- Continue onwards and hit, potentially fatally, the young child
Technology is only one part of the question when you are trying to get autonomous vehicles on the road. Until this question can be answered, and agreed, the world will not be comfortable handing over control to a machine.
Khosrowshahi might be pinning his hopes on Uber profitability on the introduction of autonomous vehicles and the removal of humans (along with their minimum wage and holiday pay requirements), but we cannot see much credibility to the timescales he has put forward.