Waymo is already testing cars on limited-but-public roads in Arizona.
Over the past ten years, deep learning — a method that uses layered machine-learning algorithms to extract structured information from massive data sets — has driven almost unthinkable progress in AI and the tech industry. It powers Google Search, the Facebook News Feed, conversational speech-to-text algorithms, and champion Go-playing systems.
That's not entirely accurate, and it's an important distinction. Google's search algorithms are hand-manipulated and optimized. More importantly, Google's - and therefore Waymo's - approach is not to throw it all at AI or deep learning, but to code a system where they can verify the ins and verify the outs. Tesla and Uber are very much trying to solve the problem through autonomy but Waymo doesn't work unless it's tethered strongly to the mothership.
From a business perspective it's the only thing that makes sense. Sell a car with AI and you've sold a car. Sell a car with Waymo SelfDrive (or whatever they'll call it) and you've sold a service. That service is a reliable monthly that you can manipulate however you want, you can change the featureset, you can update at will.
It's important to point out that AI is gonna hit the wall. But it's negligent to suggest that it's the only approach in the space.