I have never heard this: "But we were promised that robotaxis would not make human driving mistakes" Please tell me who said this? I have honestly never heard it claimed by the self driving car guys. What I heard was that they would make fewer human driving mistakes, and that overall they would be safer. I find this complaint making up a lie to make a fallacious case. Sadly this kind of 'exaggeration' just sillifies the laws. Do we make it illegal to make any mistake a human might make? (Drunk or a rat shorting a wire?). In any event, the second measure is measurable: "Overall safer." Does Robotaxi work like uber in terms of getting a ride? With uber you have to say where you get picked up and where you are going. I would assume, like uber, sometimes you will be disappointed with no uber because the no uber driver will take the offer. I spent a lot of time on Ubers (or Lyfts) and learned where and why to have a friend drive me. (Never try to cross the river in Boston city proper to Cambridge or vice versa). Teslas were popular and a good FSD guy (/girl) was pretty amazing at how they could anticipate where the automation was likely to make mistakes. I am certain these systems must have a way to risk-evaluate routes. I would love to be able to order a robotaxi if it can find a safe(r) route. I know when I am driving, I often take the longer, less stressy, route just to avoid the possibility of something bad happening (like in Atlanta, DC, or Southern California). I can't wait to have a voice Grok for my robotaxi so I can ask questions and give advice. (Hi Phil).
We should hold the manufacturer liable for harm caused by a robotaxi making a mistake which, if it were a human driver, we would consider negligent driving. That is not a standard of perfection. While that is not an optimum safety standard, it is the only one that will not shatter the court system in practice. See my work with W. Widen on that topic (in my CMU paper & talk lists)
Here is a starting point for the promises that have been made. The reality is that people make some types of mistakes, robotaxis make other types of mistakes, and surprisingly many of those mistakes have similar symptoms even if different root causes.
Reuters has an article that takes a look at various incidents that have been reported with Tesla robotaxi driving behavior in Austin:
https://www.reuters.com/business/autos-transportation/teslas-robotaxi-peppered-with-driving-mistakes-texas-tests-2025-06-25/
I have never heard this: "But we were promised that robotaxis would not make human driving mistakes" Please tell me who said this? I have honestly never heard it claimed by the self driving car guys. What I heard was that they would make fewer human driving mistakes, and that overall they would be safer. I find this complaint making up a lie to make a fallacious case. Sadly this kind of 'exaggeration' just sillifies the laws. Do we make it illegal to make any mistake a human might make? (Drunk or a rat shorting a wire?). In any event, the second measure is measurable: "Overall safer." Does Robotaxi work like uber in terms of getting a ride? With uber you have to say where you get picked up and where you are going. I would assume, like uber, sometimes you will be disappointed with no uber because the no uber driver will take the offer. I spent a lot of time on Ubers (or Lyfts) and learned where and why to have a friend drive me. (Never try to cross the river in Boston city proper to Cambridge or vice versa). Teslas were popular and a good FSD guy (/girl) was pretty amazing at how they could anticipate where the automation was likely to make mistakes. I am certain these systems must have a way to risk-evaluate routes. I would love to be able to order a robotaxi if it can find a safe(r) route. I know when I am driving, I often take the longer, less stressy, route just to avoid the possibility of something bad happening (like in Atlanta, DC, or Southern California). I can't wait to have a voice Grok for my robotaxi so I can ask questions and give advice. (Hi Phil).
We should hold the manufacturer liable for harm caused by a robotaxi making a mistake which, if it were a human driver, we would consider negligent driving. That is not a standard of perfection. While that is not an optimum safety standard, it is the only one that will not shatter the court system in practice. See my work with W. Widen on that topic (in my CMU paper & talk lists)
Here is a starting point for the promises that have been made. The reality is that people make some types of mistakes, robotaxis make other types of mistakes, and surprisingly many of those mistakes have similar symptoms even if different root causes.
https://youtu.be/pYb4X5aJhgU&t=162