:date: 2024-12-11 13:47 :tags:
The AP reports that GM to retreat from robotaxis and stop funding its Cruise autonomous vehicle unit. While the headline says all you need to know, this bit is also interesting: "Since GM bought a controlling stake in Cruise for $581 million in 2016, the robotaxi service piled up more than $10 billion in operating losses while bringing in less than $500 million in revenue..."
Ok, fine. We knew it was an R&D project really, so losing astronomical sums of cash was expected. What is interesting is the bigger picture insight into what the car industry thinks of automous vehicles. In 2016, sure, GM and the rest had no idea just how magical the tech wizard magic was. They could read sci-fi stories like anyone else and if the magic really was powerful, they had a legitimate FOMO.
Compare GM's apparent feelings about the future of driverless cars with Waymo's. A month ago The AP reported Waymo’s robotaxis now open to anyone who wants a driverless ride in Los Angeles. And a few days ago, they reported Waymo unveils plan to bring its robotaxi service to Miami. That article also reminds us: "Having periodically promised a fleet of Tesla robotaxis for nearly a decade, Tesla CEO Elon Musk renewed the pledge again in October when he predicted the electric carmaker’s 'Cybercabs' will be on U.S. roads in 2026."
Let's ignore Musk for the moment and consider what these conflicting developments mean. I propose it is one of these.
One possibility is that GM is managed by fools who have come so close to having the most transformative product since the automobile itself and have stupidly thrown it away. Maybe their bad luck with a notorious non-fatal pedestrian injuring accident and clumsy handling of that incident left them too depleted to carry on (report here). That's basically what happened to Uber (my analysis of that). But that's roughly the first possibility: GM being generally incompetent at transportation.
The other possibility is that Waymo's hype and limited fair weather "robotaxi" service is not as ready to take over the streets as the positive sounding deployment news suggests. For example, none of the Waymo deployment locations (SF, Phoenix, Austin, Miami) ever see snow or ice. And, as Waymo's own web site says, "Much like phone-a-friend, when the Waymo vehicle encounters a particular situation on the road, the autonomous driver can reach out to a human fleet response agent for additional information to contextualize its environment." In other words these vehicles have a non-autonomous fallback mechanism.
My opinion on the matter has remained unchanged since 2016 — I think that getting computers to drive cars like humans in a human environment is tantamount to designing a computer that can do any particular thing a human can do. And for many things — especially those spatial control things that animals interacting with each other are good at — the technology is just not there.
So are Waymo and Tesla right (optimism), or GM and Uber (pessimism)? It's possible that if Waymo can deploy enough constrained examples of mostly autonomous fair weather driving that it could conceivably move the general expectations in their favor. This would mean that finally dedicated lanes/infrastructure/rules/planning would be a plausible political reality. Of course if they could really saturate a market somehow past the critical threshold, the problem (which is and always has been idiot human drivers) cures itself.
I don't really know what the idiot humans writ large are going to decide about this. What I can advise is that you should view the situation with more pessimism than Waymo/Tesla. But you should also be more optimistic about autonomous vehicles succeeding than GM/Uber.