Nearly 20 years ago, I predicted that no street-legal, fully autonomous automobiles would be operating on the streets of American cities for the foreseeable future. That was received with scoffing by some of my readers. If Waymo is to be believed, nearly 20 years have now elapsed and the fine print tells the real story:
The Waymo Driver autonomously navigates tens of thousands of rider-only miles across San Francisco, Phoenix, Los Angeles, and Austin daily. It can navigate common scenarios, like adhering to a crossing guard directing traffic, as well as more unique interactions like avoiding a swerving vehicle. As the Waymo Driver travels across town, it might contact fleet response for additional help.
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. The Waymo Driver does not rely solely on the inputs it receives from the fleet response agent and it is in control of the vehicle at all times. As the Waymo Driver waits for input from fleet response, and even after receiving it, the Waymo Driver continues using available information to inform its decisions. This is important because, given the dynamic conditions on the road, the environment around the car can change, which either remedies the situation or influences how the Waymo Driver should proceed. In fact, the vast majority of such situations are resolved, without assistance, by the Waymo Driver.
or, in other words, Waymo’s robotaxi is not fully autonomous. If the system requires human judgment to resolve edge cases in real time, it is not fully autonomous. It is largely autonomous and that is a development to be applauded.
The irony of this is that what they’re actually doing is the right way: autonomous with human oversight. For decades I have said that I would joyfully accept fully autonomous vehicles in our streets if strict liability applied. That would align incentives properly. Waymo’s mistake is in advertising their robotaxis as “fully autonomous” rather than what they actually are.
The main open question is whether you trust operators potentially thousands of miles away to get Waymos in San Francisco out of tricky situations.
I think you are mistaking “always autonomous” with “high autonomy”. Using the accepted standard for autonomy by industry and government (SAE); Waymo is “level 4”.
Level 4 is defined as “vehicles that can handle all aspects of driving, monitor the environment, and navigate safely without human intervention, but only within a specific, restricted geographical area or under certain conditions (geofencing) … If a system failure occurs or the vehicle approaches the edge of its operational area, the vehicle can independently navigate to a “minimal risk condition” (e.g., pulling over) without human input”.
Level 5 is defined as “where the vehicle’s system handles all driving tasks under all conditions, with no human interaction or steering wheel required”.
My understanding is the intervention rate for Waymo’s is very. low — much lower then a human driving a car would require.
Based on Waymo’s post, it does not meet Level 4.
The distinction between “very low” and “none” is non-trivial.
The autonomy is achieved by constraining the problem not solving it universally. That is contrary to ordinary usage.
My understanding is based on Waymo’s data submitted to the state of California, the intervention rate is about 1 per 10000-20000 miles. Given the average ride is about 10 miles, that’s about 1 intervention per 1000-2000 rides. And most of the interventions are non-critical (e.g. passenger passed out).
Waymo’s are not constrained as much as you think. But of course if you want to run a transportation system you have to have backups and backups to backups, and humans in the loop. AI haven’t learned to be as adaptable to humans to the unexpected yet.
This I think is a fair take. https://thelastdriverlicenseholder.com/2026/02/09/are-waymos-remote-controlled-or-not-the-answer-is-no/
My sense is that from the POV of the consumer CO is more correct. The cars are already safer, in the areas they are testing, than humans when driving. The cars have bettered human error rates. However, the Waymo cars are aiming for zero error rates as they train. They would already be autonomous even by Dave’s dictionary definition if the standard was meeting or exceeding human performance.
I guess the good news is that the cars can be trained for those remains few complications they have not seen yet as they encounter them. Humans on the other hand are going to be stuck with an unavildoable amount of bad judgment, inattention and substance abuse.
Steve
OT- Article you might like. Turns out that states like California and Illinois, compared with other states, arent losing people at a high rate. Rather, people are moving into them at a lower rate.
“Let my trusty spreadsheet explain, using the just-released Census Bureau’s state-to-state migration stats for 2024 that track relocation patterns among the population ages 1 year and older. Comparing that year with 2021-23 averages adds perspective.
These fresh demographic figures show California lost 661,000 residents to other states in 2024, the most exits in the nation. By the way, those departures are 16% below the 2021-23 pace.
But before you shout “exodus,” note that Florida was No. 2 at 506,000, followed by Texas at 483,500.
There’s a theme here. These economic rivals are also the nation’s most populous states – California with 39 million, Texas with 31 million and Florida with 23 million.
In a nation where 7 million people got a new home state in 2024, it’s not totally unexpected that big states have more interstate moves. By the way, moving to another state is down 9% vs. 2021-23.
So ponder these exits as a share of all residents. Consider this departure rate as the odds that you knew someone who left for another state.
Note that 2.1% of Americans changed states in 2024. That same year, 1.7% of Californians left for other parts of the U.S. That below-average departure rate is also down from California’s 2% departure rate in 2021-23.
California is among the states with the most loyal residents. For 2024, only Michigan (1.3%), Ohio (1.5%), and Texas (1.6%) had smaller departure rates.
https://archive.ph/885lC#selection-2083.0-2115.152
I’m well aware of California’s “stickiness”. My in-laws are all native Californians. It’s a quality that California does not share with Illinois. That’s why I have insisted for 30 years that Illinois needs to be a place where businesses can prosper—it’s a place people come to work.
The cited article is somewhat misleading. The issue isn’t “departure rate”. It’s net domestic migration. California has depended on net positive domestic inmigration for the last century; that isn’t happening now. Illinois hasn’t had net positive inmigration in decades.