I believe there’s a missing nuance in why people oppose AVs : there’s a perception bias of giving up liberty. Roads are still human spaces, where human agency and autonomy still precedes algorithmic rules and order. Roads still constitute a significant portion of our perception of daily space, livelihood, the means to get to family outings or an evening stroll to a nearby dine out. With AVs, there’s an artificial sense of giving up these spaces to automations and procedure. There’s a fair amount of masked conflict(however true or false they are factually) that isn’t being communicated or probably recognised. Effectively, the opposition is a critique of an algorithmic enclosure to what is fundamentally a very human, public commons. Instead of making this a nudge based shift, there’s a strong sense that it’s an imposed change that puts a disproportionate bias on eliminating the human agency here.
How Self-Driving Cars will Destroy Cities (and what to do about it)
by Not Just Bikes
It articulates this point to a great length. I have agreed with you on almost everything you wrote about but I am definitely very much anti AV and I would like to see them restricted as much as possible.
I gotta say I disagree with most of the points in this video. It's mostly referring to a few random studies from orgs that don't seem to have much influence over AVs or cities. The orgs do studies like "Yeah giving a road over entirely to AVs would be good for the AVs" and the video seems to imply "That's what's going to happen then." It's implying city planners will just completely give over cities to AVs because their lobbying power will be so strong, including getting rid of traffic lights so pedestrians can't cross streets. That's a wild over-extrapolation.
Do you know any organization, planner, urbanist, transport advocate who is both progressive and sees AVs as something that largely aligns with progressive goals, rather than a problem to be managed? By progressive I mean prioritizing people over cars, aiming for equitable, sustainable, human-scaled communities. Even if AVs proponents advertise certain elements of this, there is little actual evidence that, thinking in actual urban outcomes, AVs meaningfully advance these goals. At least I am not aware of anyone in this space who isn't treating AVs with at least deep distrust (the video I recommended would fall more on the "hate" side of distrust, which I wholeheartedly share). And with good reason, as induced demand and sprawl, infrastructure lock-in etc are quite real risks.
The only people unironically enthusiastic about AV tech seem to be tech bros, investors, industry, and certain policymakers who think of it as some symbol of tech supremacy. (Again please correct me as I probably live in a progressive urbanist bubble.)
I don't know much about the planning community so will need to dig in more. I'd recommend the part of the city council video where disability advocates start speaking. There is a lot of general interest in this beyond "techbros" https://www.youtube.com/watch?v=mzMfEsPJn0Q&t=5799s
I guess more broadly, do you worry that the progressive planning community might have been negatively polarized against AVs because they associate them with big tech more broadly? Like if I could snap my fingers and replace every car in America with an AV, I'd do it without thinking about it. I think not having 40k people die each year in accidents also needs to be factored into progressive goals. If the lock-in problem is coming from everyone wanting to use them because they work so well I think we need to step back and look at the trade-offs.
That’s really interesting. I’ve been thinking of roads as one of the places where embodied AI (in this case, self-driving cars) is likely to work best because they are *already* an inhuman space where we have given up our freedom. You can’t exist in a street or road unless you are in one of a few vehicle types, and move whenever the light turns green, and behave in very simple and predictable ways. I’ve been imagining that a growth of autonomous vehicles would take away some freedoms from people using cars, but give back some freedoms to people on foot or bike, because autonomous vehicles are much more able to deal safely with people on foot or bike than people in cars can. But I suppose it definitely isn’t obvious which direction it will go.
One of my points is about the constraint’s mechanism and how it gets perceived en masse. Today most constraints are human and contestable after the fact. AV scaling shifts them to code that is pre‑emptive, private, and hard to contest. That shift can enclose a public commons unless we design for the opposite. Hence - the nudging has to be centered around the fact that AVs can give freedoms to walkers and cyclists, but not by default. It takes explicit guardrails. And this unknown , codified guardrails eliminate the human ability to contest.
If we can learn anything from history is that the impact of technology never works out as advertised. General Motors promised "Futurama" but instead we got stroads and car dependency.
I really like the book by Peter Norton (not the programmer) Fighting Traffic as it gives a historical perspective of how cars took over the physical and social space. Anyone who thinks that AVs would not further take away from what little is left from "roads as human spaces" is either hopelessly naive or profits immensely from further takeover.
I live in Europe and I want better public transport and pedestrian & bike infrastructure. I don't want AVs as I don't believe for a second that they would lead to fewer cars overall, the comparison to car rental is apt as I don't think cheaper car rental would solve any real traffic problem. Those I think are entirely political and don't actually depend on magic new tech.
Great article, thanks Andy! It’s helpful to see all the arguments laid out clearly in one place.
One small point: while ride-share jobs may not be great jobs, the fact that many people choose them suggests they’re still the best available option for those workers. In my experience, a lot of ride share drivers tend to be low-skilled immigrants with limited opportunities to begin with. Reducing the number of such jobs might therefore harm the people who rely on them. That said, I don't think this is reason enough to ban AVs, far from it, but it seems worth noting.
I absolutely agree. Here are actions I've written connecting you with your city leaders and sending prewritten letters advocating against banning autonomous vehicles!
Great post, but I think you overstate how bad ride share driver jobs are. A ride share driver is in many respects, their own boss. They choose when they work, they choose how hard they work, and they choose how much they work. They can take a week off if they like, or just work one day a week if they like. Of course they still have to work within the confines set by uber/lyft, which means they have to accept the opaque prices uber/lyft, they have little job security, and they have to obey new rule changes from uber/lyft, but it's really not that dissimilar from, say, a freelancer having to cater to the whims of a client.
Ride share drivers trade job security and transparency for flexibility. A perfectly fine trade off for many people.
Great post. What do you think of Brad Templeton's case (from a decade ago) that ultralight single-seat AVs would be more environmentally friendly (and convenient!) than bulky public transport? See: https://www.templetons.com/brad/robocars/future-transit.html
I’ve generally been convinced by Jarrett Walker’s takes that for places where congestion is ever an issue, bigger vehicles will be a more efficient use of space than personal vehicles. It’s going to be difficult for a bunch of small vehicles to use less energy than one big vehicle, or less space, even if you’re talking about an off-peak bus with just two or three passengers on it. (Though the wear on the roadway does seem to be something that pushes towards more small vehicles rather than fewer large ones.)
I think there is a common tendency in progressive spaces to distrust technology. Whenever wealthy people use technology to lower wages (and they do this a lot), the tendency is to blame the technology itself rather than the wealthy people. But in fact technology is the only thing that has ever improved the living standards of ordinary people.
I feel very alienated by leftists around me who proudly proclaim they oppose AVs and AI. And before that they opposed the Internet, chain stores, chain restaurants, pretty much anything new. Strangely, almost none avoid social media, though they blame it for everything.
There are limitations for the reliability of data on safety of AVs. I view it more or less promotional/marketing data and narrative device to sell AVs to the public and regulators, rather than hard/scientific fact and proven safety outcome.
The actual data is opaque, proprietary, released selectively e.g. we don't know how often human operators need to intervene etc. (eg. they self-report disengagements, but there's no standard definition or audit). Also, claims like "100 million miles without a fatality" aren't terribly statistically meaningful without the context (the US has about 1 road fatality per 100 million miles driven). AVs operate only in specific, well-mapped areas under limited conditions (so there's self selection bias when comparing safety stats).
Overall I think there's an illusion of scale and abundance of evidence on safety, when in reality it's not really there yet.
Well yeah we have limited data right now, but I’d really strongly recommend the Understanding AI post I linked in that section. Third party data just makes it look pretty likely that Waymo hasn’t been at fault for a single accident yet. That’s all we have to go on, but imo is a pretty powerful data point
I understand reasons for cautious optimism, but if you're honest the emphasis should be on "cautious". Eg. even waymo acknowledges that there isn't enough data yet for statistically meaningful statements on fatalities, comparison to human drivers has selection bias (AV companies select the easiest places and situations to drive) and so on.
Fundamentally it's a traffic planning issue of mobility systems, not a tech-innovation issue. Perfect automation in a bad street system will be... bad. And I don't believe for a second that safety is an *objective* of developing AVs. If human safety was the true objective, it's possible to achieve that at a fraction of the cost with proven methods not dependent on not yet existing tech (but it involves reducing the number and speed of cars). Safety isn't an objective, rather a constraint. We will see about the true objective when the public is convinced that they are "safe enough".
You haven’t presented any economic plans for displaced workers. Uber driving is a last resort for the laid off and long term unemployed. I don’t know if you’re a paid shill or just a misanthrope.
You’re calling me a paid shill and a misanthrope because I don’t like Uber? Would love to take this back in time to when Uber was starting off to show people this. To put it mildly people had different opinions. The idea that city governments should ban whole new vehicles to maximize Uber jobs would've been seen as the misanthropic shill position. My economic plan is "don't ban Uber-like jobs from existing" which many opponents of AVs are also trying to do.
Andy, how do you deal with the emotional elements of any contentious debate? The points you raised are logical and practical but embed in them assumptions that people only care for safety or efficiency. As Nitthin mentioned, a lot of people feel like they lose something e.g. freedom because of this change. Or the idea of a big corporation doing something inherently sparks suspicion or they may not be convinced that these are 'safer'. I'm sure you've come across this in the environmental debate. I'd love to have productive conversations with people who start of wary.
I believe there’s a missing nuance in why people oppose AVs : there’s a perception bias of giving up liberty. Roads are still human spaces, where human agency and autonomy still precedes algorithmic rules and order. Roads still constitute a significant portion of our perception of daily space, livelihood, the means to get to family outings or an evening stroll to a nearby dine out. With AVs, there’s an artificial sense of giving up these spaces to automations and procedure. There’s a fair amount of masked conflict(however true or false they are factually) that isn’t being communicated or probably recognised. Effectively, the opposition is a critique of an algorithmic enclosure to what is fundamentally a very human, public commons. Instead of making this a nudge based shift, there’s a strong sense that it’s an imposed change that puts a disproportionate bias on eliminating the human agency here.
Great. Will respond to this in the post tomorrow when I have time. Silly I didn't think of it, thank you for catching and writing it out
If you didn't already, you could refer to:
How Self-Driving Cars will Destroy Cities (and what to do about it)
by Not Just Bikes
It articulates this point to a great length. I have agreed with you on almost everything you wrote about but I am definitely very much anti AV and I would like to see them restricted as much as possible.
I gotta say I disagree with most of the points in this video. It's mostly referring to a few random studies from orgs that don't seem to have much influence over AVs or cities. The orgs do studies like "Yeah giving a road over entirely to AVs would be good for the AVs" and the video seems to imply "That's what's going to happen then." It's implying city planners will just completely give over cities to AVs because their lobbying power will be so strong, including getting rid of traffic lights so pedestrians can't cross streets. That's a wild over-extrapolation.
Do you know any organization, planner, urbanist, transport advocate who is both progressive and sees AVs as something that largely aligns with progressive goals, rather than a problem to be managed? By progressive I mean prioritizing people over cars, aiming for equitable, sustainable, human-scaled communities. Even if AVs proponents advertise certain elements of this, there is little actual evidence that, thinking in actual urban outcomes, AVs meaningfully advance these goals. At least I am not aware of anyone in this space who isn't treating AVs with at least deep distrust (the video I recommended would fall more on the "hate" side of distrust, which I wholeheartedly share). And with good reason, as induced demand and sprawl, infrastructure lock-in etc are quite real risks.
The only people unironically enthusiastic about AV tech seem to be tech bros, investors, industry, and certain policymakers who think of it as some symbol of tech supremacy. (Again please correct me as I probably live in a progressive urbanist bubble.)
I don't know much about the planning community so will need to dig in more. I'd recommend the part of the city council video where disability advocates start speaking. There is a lot of general interest in this beyond "techbros" https://www.youtube.com/watch?v=mzMfEsPJn0Q&t=5799s
I guess more broadly, do you worry that the progressive planning community might have been negatively polarized against AVs because they associate them with big tech more broadly? Like if I could snap my fingers and replace every car in America with an AV, I'd do it without thinking about it. I think not having 40k people die each year in accidents also needs to be factored into progressive goals. If the lock-in problem is coming from everyone wanting to use them because they work so well I think we need to step back and look at the trade-offs.
That’s really interesting. I’ve been thinking of roads as one of the places where embodied AI (in this case, self-driving cars) is likely to work best because they are *already* an inhuman space where we have given up our freedom. You can’t exist in a street or road unless you are in one of a few vehicle types, and move whenever the light turns green, and behave in very simple and predictable ways. I’ve been imagining that a growth of autonomous vehicles would take away some freedoms from people using cars, but give back some freedoms to people on foot or bike, because autonomous vehicles are much more able to deal safely with people on foot or bike than people in cars can. But I suppose it definitely isn’t obvious which direction it will go.
One of my points is about the constraint’s mechanism and how it gets perceived en masse. Today most constraints are human and contestable after the fact. AV scaling shifts them to code that is pre‑emptive, private, and hard to contest. That shift can enclose a public commons unless we design for the opposite. Hence - the nudging has to be centered around the fact that AVs can give freedoms to walkers and cyclists, but not by default. It takes explicit guardrails. And this unknown , codified guardrails eliminate the human ability to contest.
That's definitely an important point for me to think more about!
If we can learn anything from history is that the impact of technology never works out as advertised. General Motors promised "Futurama" but instead we got stroads and car dependency.
I really like the book by Peter Norton (not the programmer) Fighting Traffic as it gives a historical perspective of how cars took over the physical and social space. Anyone who thinks that AVs would not further take away from what little is left from "roads as human spaces" is either hopelessly naive or profits immensely from further takeover.
I live in Europe and I want better public transport and pedestrian & bike infrastructure. I don't want AVs as I don't believe for a second that they would lead to fewer cars overall, the comparison to car rental is apt as I don't think cheaper car rental would solve any real traffic problem. Those I think are entirely political and don't actually depend on magic new tech.
Here's what autonomous vehicles do NOT do
Drive drunk or high
Road rage
Flee accidents
Fall asleep at the wheel
Look at at their phone while driving
Ignore traffic laws
Tailgate , Rush, or Rubberneck
Get angry at other drivers
Do donuts on the freeway as recreation
Well for sure - it's because AVs are predominantly just software.
But here are things AVs definitely do:
- go wrong way into oncoming traffic
- trap passengers and drive them in circles
- honk and circle all night in a parking lot
- randomly crash into poles
- get disabled by a traffic cone
- hit other robots
- get recalled
- get confused by obvious police hand signals
- cut each other off
- gridlock each other
fine i wont.
Great article, thanks Andy! It’s helpful to see all the arguments laid out clearly in one place.
One small point: while ride-share jobs may not be great jobs, the fact that many people choose them suggests they’re still the best available option for those workers. In my experience, a lot of ride share drivers tend to be low-skilled immigrants with limited opportunities to begin with. Reducing the number of such jobs might therefore harm the people who rely on them. That said, I don't think this is reason enough to ban AVs, far from it, but it seems worth noting.
Will add a point on this, thanks Maxime!
I absolutely agree. Here are actions I've written connecting you with your city leaders and sending prewritten letters advocating against banning autonomous vehicles!
https://climateactapp.substack.com/p/your-daily-dose-of-climate-hope-july-a94?utm_source=publication-search
https://climateactapp.substack.com/p/your-daily-dose-of-climate-hope-september-d21?utm_source=publication-search
Great post, but I think you overstate how bad ride share driver jobs are. A ride share driver is in many respects, their own boss. They choose when they work, they choose how hard they work, and they choose how much they work. They can take a week off if they like, or just work one day a week if they like. Of course they still have to work within the confines set by uber/lyft, which means they have to accept the opaque prices uber/lyft, they have little job security, and they have to obey new rule changes from uber/lyft, but it's really not that dissimilar from, say, a freelancer having to cater to the whims of a client.
Ride share drivers trade job security and transparency for flexibility. A perfectly fine trade off for many people.
Good call, I might circle back and edit. Thanks!
Great post. What do you think of Brad Templeton's case (from a decade ago) that ultralight single-seat AVs would be more environmentally friendly (and convenient!) than bulky public transport? See: https://www.templetons.com/brad/robocars/future-transit.html
I’ve generally been convinced by Jarrett Walker’s takes that for places where congestion is ever an issue, bigger vehicles will be a more efficient use of space than personal vehicles. It’s going to be difficult for a bunch of small vehicles to use less energy than one big vehicle, or less space, even if you’re talking about an off-peak bus with just two or three passengers on it. (Though the wear on the roadway does seem to be something that pushes towards more small vehicles rather than fewer large ones.)
https://humantransit.org/category/automation
"Every hour people drive Ubers is an hour they're not out on the street, doing drugs and committing crime." https://www.astralcodexten.com/p/highlights-from-the-comments-on-missing
The main difference is that your points rarely happen at my points never happen.
Excellent post.
I think there is a common tendency in progressive spaces to distrust technology. Whenever wealthy people use technology to lower wages (and they do this a lot), the tendency is to blame the technology itself rather than the wealthy people. But in fact technology is the only thing that has ever improved the living standards of ordinary people.
I feel very alienated by leftists around me who proudly proclaim they oppose AVs and AI. And before that they opposed the Internet, chain stores, chain restaurants, pretty much anything new. Strangely, almost none avoid social media, though they blame it for everything.
One issue with your safety argument:
There are limitations for the reliability of data on safety of AVs. I view it more or less promotional/marketing data and narrative device to sell AVs to the public and regulators, rather than hard/scientific fact and proven safety outcome.
The actual data is opaque, proprietary, released selectively e.g. we don't know how often human operators need to intervene etc. (eg. they self-report disengagements, but there's no standard definition or audit). Also, claims like "100 million miles without a fatality" aren't terribly statistically meaningful without the context (the US has about 1 road fatality per 100 million miles driven). AVs operate only in specific, well-mapped areas under limited conditions (so there's self selection bias when comparing safety stats).
Overall I think there's an illusion of scale and abundance of evidence on safety, when in reality it's not really there yet.
Well yeah we have limited data right now, but I’d really strongly recommend the Understanding AI post I linked in that section. Third party data just makes it look pretty likely that Waymo hasn’t been at fault for a single accident yet. That’s all we have to go on, but imo is a pretty powerful data point
I understand reasons for cautious optimism, but if you're honest the emphasis should be on "cautious". Eg. even waymo acknowledges that there isn't enough data yet for statistically meaningful statements on fatalities, comparison to human drivers has selection bias (AV companies select the easiest places and situations to drive) and so on.
Fundamentally it's a traffic planning issue of mobility systems, not a tech-innovation issue. Perfect automation in a bad street system will be... bad. And I don't believe for a second that safety is an *objective* of developing AVs. If human safety was the true objective, it's possible to achieve that at a fraction of the cost with proven methods not dependent on not yet existing tech (but it involves reducing the number and speed of cars). Safety isn't an objective, rather a constraint. We will see about the true objective when the public is convinced that they are "safe enough".
You haven’t presented any economic plans for displaced workers. Uber driving is a last resort for the laid off and long term unemployed. I don’t know if you’re a paid shill or just a misanthrope.
You’re calling me a paid shill and a misanthrope because I don’t like Uber? Would love to take this back in time to when Uber was starting off to show people this. To put it mildly people had different opinions. The idea that city governments should ban whole new vehicles to maximize Uber jobs would've been seen as the misanthropic shill position. My economic plan is "don't ban Uber-like jobs from existing" which many opponents of AVs are also trying to do.
Are Uber and Lyft drivers having trouble in Los Angeles and San Francisco?
Andy, how do you deal with the emotional elements of any contentious debate? The points you raised are logical and practical but embed in them assumptions that people only care for safety or efficiency. As Nitthin mentioned, a lot of people feel like they lose something e.g. freedom because of this change. Or the idea of a big corporation doing something inherently sparks suspicion or they may not be convinced that these are 'safer'. I'm sure you've come across this in the environmental debate. I'd love to have productive conversations with people who start of wary.