Uber CEO says Uber and Alphabet are discussing working together on self-driving

Hell yes. A lot of hype has been sold at a dear price, driven in no small part by Uber’s desperation to find a profitable business model through automation. The reality is that critical edge cases are hard, and no one seems to know how to fix them. Two factors that I believe will dictate “decades away” or as you rightly say, someday, but not today, probably a loooong time from now. First, these crashes are not going to stop as long as SDVs are on public roads, and that includes the bad joke that is Tesla AP. That kind of thing is just too juicy for the media not to latch on to, and too scary to the average person to ignore. As a thought experiment, imagine if Uber has mowed down a well-off white kid instead of someone they could dismiss as a “formerly homeless woman.”

Second, the tech itself is hard and may be on shaky ground to begin with. The assumption that with enough training AI can adapt to the point of “better than human” is yet to be borne out in reality. And yet, the billions have already been invested, so a lot of people have a lot on the line to make some kind of MVP. So “millions of miles driven” gets bandied about, but it’s not miles in torrential rain, or terrible roads, or snow and ice. Who cares how many miles Waymo can drive on a sunny day in California? Get back to me when they can do the same on a back road in Maine, in the winter.

IMO this creates a feedback loop of hype and investment until the bottom drops out. Then it won’t just be a technological problem, but political, legal, and shrouded in “winter” as investors once burned will be reluctant to wade in again. To the people who see the billions invested as evidence of promise in and of itself, I say Theranos... Juicero...

In the case of the auto manufacturers, subsets of autonomous driving are interesting. If a car can truly be hands and eyes-off on highways (and the tech for that is fairly close), that's a product that almost sells itself.

I'm much more skeptical about driverless cars doing arbitrary pickups and dropoffs. I do think that anyone who is investing on the assumption that an Uber will be able to offer that in a few years is going to be disappointed. It will happen but not quickly with a few caveats like maybe fixed locations can be established to simplify the problem.