Anyone rooting against self-driving cars is cheering for tens of thousands of deaths, year after year. We shouldn’t be burning self-driving cars in the streets. We should be celebrating them.
1. Luxury belief. The people advocating most loudly for implementing these are the ones that do not live in the places these are tested. You’re welcome to donate your neighborhood and children to the testing curve that goes with implementation of any new technology. After all if your child dies you can console yourself that, statistically, it’s for the greater good.
2. Lack of data. These have never been deployed at scale. Saying they are safer than human drivers (implication: at scale) is like saying a barely tested vaccine is safe and effective. Maybe, maybe not. This claim is just scientifically invalid.
3. Liability matters. Right now at least there is the threat of prosecution for dangerous driving. Which court, exactly, is equipped to comb through billions of lines of code to discern who messed up? As far as I know there’s never been an autonomous system of any kind holding millions of lives all at once in any given moment, and what the incentive structures look like when liability is decoupled from individual actors. And are you sure that amongst tens of thousands of engineers there will never be one horrifically stupid or bad actor?
4. Techno utopians remain shockingly ignorant that technologies never develop as planned. Look at the horror that is too often Facebook. Or smartphones on kids. Technologies’ second, third, fourth order etc effects are inherently unpredictable. The creators are never held accountable.
5. Individuals have autonomy. In contrast, a massive system is different and susceptible to hackers beyond individual control. Generally people don’t get dismembered when their bank account is hacked. How about a hack with hundreds of thousands of drivers all doing 65 mph?
6. I’ve said the solution to this for years is not to make Americans more sedentary by car expansion but restructure cities to be human and child centric. Most Americans live in cities. Many cities in the world have solved this problem already with great urban planning and public transport. Cars are a bad idea for a lot more than road fatalities, in many ways they’ve diminished the beauty and livability of American cities, health, and human relationships.
7. Bad government actors. We’ve seen totalitarian overreach in many areas and it’s even worse in Canada. If governments control social media accounts and cancel bank accounts for political persecution, what exactly does a protest look like when those same corrupt government officials are installed at Waymo just like they were at Twitter?
My point: human freedom isn’t possible when large opaque systems can in one moment derail someone’s intentions or life. There are many other highly successful interventions already in use to reduce traffic deaths that don’t have any of the above risks. Even with a car centric city like Portland they’ve reduced many of the speed limits to 20mph, for example. AI driving is a different animal from past technologies in many dimensions and should therefore be in its own category and not naively equated to the rollouts of past technologies.
Re: (7) .... imagine the future in which all travel is by automated car and no personal vehicle ownership is permitted (the future they want). They can make you a virtual prisoner by shutting off your vehicle access. They can prevent you from traveling certain places at certain times, or if an event is happening.
Your point 6 sounds a lot like the 15-minute cities that Globalists want to cram us into.
Part of the reason that public transit works better in Europe is because the much denser population can sustain it. That is not true of the vast majority of American cities. I've tried to get around mid-sized American cities with public transit, and it's a much different (and far more inconvenient) experience than my experience with public transit in European cities.
The reality is that most American cities lack the population density to make public transit routes that go where a lot of people need to go practicable. One *could* design the necessary routes and scheduling, but many of the buses would be traveling empty or nearly so too often to justify the expense involved.
I would dearly love to live in a 15-minute city, which was effectively made illegal in most U.S. jurisdictions for the past six or seven decades.
I've gotten as close as I can to it: I can commute to work by train, walking to and from the station on both ends. The one thing I can't do on foot is to shop for things I need, not even groceries. I mean, there are lots of twee boutiques in town, but man cannot live by sustainably-sourced hand-knit wool sweaters alone.
When I lived in France, in the northern suburbs of Paris, in the early 1990s, one of the very best things about it was that there was a bakery and a small grocery, side by side, directly across the street from my apartment . . . and a commuter rail station just down the street.
That was a good and decent way for a human being to live, and it didn't even have to be forced by (allegedly) tyrannical planners. All you have to do is not make that stuff illegal, not force zoning and planning codes that separate different land uses from one another, move things farther apart, and push them farther back from the street. In fact, the most discordant and dysfunctional parts of the northern suburbs of Paris are around the big ugly tower blocks in the style of le Corbusier, who wanted to impose his own modernist, car-centered vision on everyone.
What I'm describing are just the natural forms human settlements took in the past, before the regimentation of zoning and the effects of the automobile.
I live in a suburban city and city buses usually are about 5% full all the while spewing diesel exhaust into the air. The city's infrastructure wasn't designed for buses so when a bust stops at a bus stop half of the bus is sticking out into the lane which backs up traffic and causes safety issues as drivers try and swerve around them.
The last time I was in SF I rode in a driverless car and it was nice. The car was clean, got to the destination in the most efficient route and was safe.
You're totally right about the density issue and I wouldn't want the 15-minute city type of central planning either, because personally I'm against any mass-scale bureaucratic utopia. But I love this topic so let me add some context to my above comments.
In effect, the present world is the fruition of a past utopia, the one envisioned by previous bureaucrats and heavily influenced by the auto lobby. The freeways carved through American cities destroyed majority black neighborhoods (and also European Ethnic neighborhoods like Greeks and Italians in Chicago, for example), not to mention many waterfronts. Definitely wasn't democratic. If I recall my history, GM deliberately dismantled LA's formerly great public transport system (and faced anti-trust lawsuits) and in New York Robert Moses built roads specifically to contain minorities in urban areas, among many examples. The car-centric city was foisted upon Americans through federal and local (zoning) policies, motivated by a variety of ideologies, and certainly not crowd sourced in a free market. Yet prior to this policy shift, urban mass transport networks (often private) thrived, decades after car ownership was widespread. I'm not an expert in these areas so I can't weigh all the factors, and certainly voluntary car adoption was one factor of how we got here, but, my point is that American cities haven't been allowed to grow organically in the first place.
Looking at the details, the reason many American cities lack density is because of explicit decades-long zoning laws banning it. This has recently changed in some places like Portland but the barrier remains still that, for example, a "cheap" $350k 1-bedroom in SF costs exactly twice that after all the bureaucrats get done charging bogus permitting fees. It can take years to develop a parcel as a result. Yet Americans want density. Some of the most expensive neighborhoods in the country are the densest. People voluntarily move to these neighborhoods and pay huge percentages of their incomes to live there. People want to live in fun places where their kids can walk around without cars driving fast. People want to be able to walk to the grocery store, not because of eco-ideology but because it's really enjoyable. Americans visit Europe by the millions just to experience this sort of thing. So in my view things like density are downstream from a policy problem which creates a density problem, and the density problem creates high speed transport infrastructure leading to an excess road fatality problem. The criticism I have of the above articles is that it just quotes one statistic (road fatalities), out of context, and then uses that to justify a large-scale transformation of our entire society. We see this pattern of argument in many areas but I'll leave that for another time. I'd like to live in a world that is neither an auto-corporation utopia nor a globalist utopia, and certainly not an AI utopia to compensate for the failures of the auto utopia. So this is what I mean by the solutions being so much simpler than all that: let Americans build out their own cities in relative freedom. Density will take time to build out but we need to take the long view.
I disagree with your assertions about car-centric cities not being "crowd sourced in the free market." While specific details of highway planning are due to bureaucrats, the demand for those highways came from car owners.
The automobile made it possible for people to live further from their workplace. It led directly to suburbs and urban sprawl. Although *you* may love cities, with all their crowding and noise (and now, increasingly, crime and filth), many people do not.
In the late 19th and early 20th century, there was a mass of population movement from the country to the city. Young people were sick of the hard work involved with farming, and cities offered many things that the country could not. But as cars were becoming more and more available, many of those people realized that high population density was not so pleasant to live in night and day. They yearned to have a bit of the country they'd grown up in, and cars made it possible to live in the country areas near the city. Housing developers met the rising demand. Suburbs were born.
High demand for expensive housing in dense urban areas is driven by people who thrive on city life, who don't mind living in apartment buildings with many (not always polite) neighbors, and who don't mind crowding. In some places (like SF) there is status associated with such neighborhoods.
But a lot of people *hate* living like that. You literally could not *pay* me in live in such conditions.
It sounds like you are angry about cities. Honestly, it would be nice for you to have thoroughly read my post before commenting as you are responding to points that I didn't make, and missing points that I did make, including regarding the history of suburbanization and individual choice. But in general I'm not interested in debating whether or not you enjoy living in cities. My point is to challenge the author's simple-minded assertion that the solution to traffic deaths is AI when there are many policy decisions that contribute to traffic deaths, particularly in sprawling urban areas where the majority of Americans live. Again, for example, if I could vote between (a) spending hundreds of billions on AI or (b) allowing individuals to build out urban density because they want to, I'd choose b.
I generally agree with your points here, except I think you're still giving too much credit and leeway to those "techno utopians" you deride. Walkable cities are fine if that's what you prefer, but not everyone wants to live in a city (some of us prefer quiet and wide-open space, farmable land, and polite people). The automobile is one of the greatest markers of freedom and autonomy, especially in a country as big and varied as the US. That's one of the reasons widespread public transport never took off. If you wake up one day and want to go to the beach (but live in OK or CO), you can start driving without having to check train schedules or anything.
As for driverless cars, getting in one means you are immediately at the mercy of a machine controlled by someone you don't know and can't see. It's nightmarish. Not only has the tech company run your background when you paid for the car service, now they have you in a speeding death trap that can take you anywhere they want. They can drive you into the ocean, into a brick wall, into oncoming traffic. At least with a human driver you have the option to try and stop them. Or, you know what, why not try this great thing - don't know if you've heard of it - but drive the car yourself. Then you can make your own decision to drive into the ocean if you'd like.
Those will automatically lock the doors and drive you to the re-indoctrination center. If you are Jewish, it will fill with the output of a small internal combustion engine.
The whole "freedom" argument for driving is very weak when one considers that driving requires vast subsidies and infrastructure and has enormous effects on those who do not drive as well as those who do. Your argument about "immediately being at the mercy of a machine controlled by someone you don't know" can be applied to airlines as well and we know they are the safest form of travel. I don't like arguments that put up a wall to protect the status quo. You cannot use a smarmy "freedom" argument to imply that there is unmanaged danger in robot vehicles while making no attempt to compare it to the unmanaged danger of the status quo. There is NOTHING that protects you and me from the driving of people who are drowsy, inebriated, high, suicidal, or just plain reckless. We are powerless against this "freedom".
As a conservative, protecting the status quo - when worth protecting - is the point. I look at the alternatives, weigh the pros and cons, and make a decision if the thing we're doing now is the best among other options. Progressives (which includes tech utopians), on the other hand, barrel through without considering the consequences of what they're destroying. The automobile was revolutionary in a way progressives can only dream of. But what it was replacing (slow/inefficient/relatively expensive forms of travel) was vastly inferior. Over time, new inventions and regulations made them more safe. And the great thing about that is, if you are so inclined, you're not forced to drive the safest car. There was an article in WSJ the other day about a guy who restored a Model A and drives it whenever he wants, even though it has little to no of the features of a modern car (he compared it to driving a tractor). And he has the freedom to do so.
I'm not even arguing that "autonomous" cars won't eventually be as safe or safer than cars are now (one of the commenters here mentioned that because there are so little data, any safety claims about them are impossible to prove). But, in the here and now, not some dreamed of utopia, and as long as I have the freedom to do so, I won't be subjecting myself to the whims of a new technology (compared to planes) that I have little to no control over.
Yes, I recognize that vast sums of federal and state dollars goes to maintaining the interstate and roads and gas stations/oil industry, but it is in the service of allowing myself and other citizens the freedom to drive wherever and whenever we like. To move from one state to another. To not be locked in our homes subsisting on virtual reality and food delivery. And yes that freedom comes with risks from stupid and reckless people, but I still prefer it to the alternative. You have the freedom to disagree.
I certainly can see why some are "late adopters" or even "never adopters" to a new technology. I will choose to an early one and that makes it better for the later ones. The crux of the disagreement is that those of us who favor robot driving will want the system to cater to our needs versus the needs of human drivers. Yes real data is necessary to prove a safety case but we must allow significant deployment of robot driving for such real data to be collected.
Does choosing "smarmy" as the adjective for your scare-quoted noun "freedom" belie a desire to have Google and/or the government, take your smarmy freedoms away in return for your supposed safety?
You are not powerless against people who are drowsy, inebriated, etc. Shame on whoever told you you are powerless. Highly skilled and situationally aware drivers are extremely good at avoiding collisions with such drivers. Not with 100% effectiveness, but a self-driving car isn't offering that either. And self-driving cars haven't been shown to be good at protecting you from any of those things. Unless, of course, you posit that they replace all human drivers.
Human agency, properly deployed, is vastly safer than anything anyone in the self-driving car industry even feigns will be on offer any time soon.
It amazes me how willing some people are to hand over to robots some of the very things that makes us human: our autonomy and our agency.
It is not a surprise that the people who are anxious for us to give our agency and autonomy to the likes of Google have been all too excited to give it to our government (see the Covid response, for example). It is not a coincidence that almost all Googlers, and government employees, are of the Democratic party. They believe Big Brother should be watching you, controlling you, thinking for you. And, of course, they imagine themselves so brilliant they should be Big Brother.
Freedom is a gift from God - you have a natural right to it in every circumstance, not because some government or company gave it to you. You don't need to trade it for security.
We are also powerless against Alphabet, Inc. which routinely abuses it's power over every freedom it is given accesss to and is the frontrunner in driverless car technology.
What does protect me from the other drivers on the road currently - however poorly - is a shitload of good laws concerning personal liability, freedom of movement, a training requirement, an insurance requirment and armed state officials enforcing these rules 24/7 with (semi) strict oversight themselves. Nowhere else in my private life am I subject to such intense scrutiny.
When I get into a Google car, I am abandoning all legal protection to their terms of service unless we put up a similar legal framework. Same as an airline, as you say.
I want driverless cars as bad as you do. I want to go places while I'm drinking without putting someone else in danger. I want my older family members to have freedom of movement. I have been injured in car wrecks, my family as well, and I've had freinds and neighbors killed.
But in the same way you wouldn't have the Sackler family that made the heroin crisis solve the pandemic (fuck, wait we did that), or the CIA determine our freedoms with the patriot act (fuck, again?) don't let Google play us for fools using fear to sell a product that needs strict oversight to prevent abuse.
With all of those safeguards you mention, there are about 40K deaths per year. I don't know anyone in my life who has been murdered, Zilch. I have had a conservation with close to a dozen people killed by a car. Is this experience any different from you? Humans driving a machine weighing over a ton at any speed they chose on their whim is literally one of the most dangerous activities that we commonly do. If I made a bet about lives killed per mile driven by human vs mile driven by robot, would you take the human side of the bet? I am so confident of the result, I would give you at least 3x odds. :-)
“You cannot use a smarmy ‘freedom’ argument to imply that there is unmanaged danger in robot vehicles while making no attempt to compare it to the unmanaged danger of the status quo.” Sir, I will have you know that this is The Free Press comment section where freedom arguments are our bread and butter. LOL
I think we agree. I'm not universally anti-car and of course there are times when cars are necessary, certainly for rural areas. Personally I love a good road trip. There are people that aren't able-bodied, etc, and they need help. Rather my points are challenging the idea that AI is the best solution to the road fatality problem. I wrote an additional comment above about that and urban density.
Funny you should mention this. Just saw that a Hawaii tourist driving a rental car and following his GPS drove over a cliff and into the ocean requiring rescue from the coastguard. Delivery trucks following GPS regularly drive up my driveway which is on the roadbed of a county highway discontinued over 50 years ago but apparently still shown on old USGS maps. Not putting my trust in GPS navigation.
The article never addresses the 226 accidents caused by driverless cars in San Francisco. Horrific accidents have involved robocars hit and runs and a pedestrian trapped under an unmoving robocar.
The author also claims that remote corrections happen efficiently, this is not always true. Ambulances have been blocked behind robocars and the robocars have mysteriously stopped causing huge traffic delays.
The anger against robocars is not a situation of angry peasants suspicious of new technology, but flaws in robocars causing accidents, blocking bike lanes, and behaving badly and dangerously. The author’s failure to address the flaws makes this a pr piece not a report.
"Accidents" are a far cry from fatalities. Car accidents are so common that no one bothers to count them. We don't even need a police report for an accident unless there is a human injury. The context here is serious human injury, not fender benders.
The NHTSA reported 392 crashes in a 10-month period involving cars with self-driving technology. Of these, there were six fatalities and five serious injuries.Sep 30, 2023 Granted that this stat may not separate self driving from cars without drivers, but there are problems.
A woman was ran over by a self driving car and it pinned her to the pavement. Emergency crews had to remove her. The self driving cars have blocked ambulances and ran over fire hoses at emergency scenes. The self driving cars have blocked traffic including bus and bike lanes for hours. While the author experienced one easy correction, his single experience does not reflect what always happens.
If you are going to write about technology, then address all the issues including its impact on SF traffic and not a puff piece that reads like pr rather than a report.
AI is probably the future of humanity, and a trillion dollar arms race favors the big dogs in the fight. It's also essential to the operation of self driving cars, which Google leads in.
Google lost $70 BILLION in market cap this week because it's AI launch couldn't make pictures of white people and called Norman Rockwell a white supremacist. Had to pull the plug after a day.
In $ value and prestige lost in 24 hours, that's the equivalent of the US Navy launching SIX new USS Ford class aircraft carriers on one day and having them all sink immediately because it would be racist to close the hatches or check the welds.... wait does Boeing make boats?
At this point I'm not sure our prospective evil tech overloards have the technical comptence to actually make the pods or fake burger bugmeat to force feed me.
I’m in favor of self driving cars and of course it’s idiotic for people to destroy them. I think it would be something of a utopia to have all of the freedom of a car without having to actually drive it.
However, as to your point 2 - right now, statistically, self driving cars are more dangerous than human driven cars on a miles driven basis.
There is 1 fatality per every ~90 million miles driven in the US. Self driving cars haven’t yet driven 90 million miles and have killed one person.
Now it’s true that the car that killed a person wasn’t a Waymo, and it’s also true that it happened several years ago.
But it’s also true that self driving cars only operate in a handful of urban locations (maybe only SF and Chandler AZ?), and not in places where ice, snow, or narrow rural roads are an issue.
Again, none of this is to denigrate self driving cars - only the ridiculous assertion that they are, right now, safer than human driven cars. They may be, but we just don’t have enough evidence yet.
All worthy points, Richard. Especially #5. I would be curious how enthusiastic our intrepid author would be sitting at the front passenger seat of a self driving commuter airplane, coming soon I hear. A hack at 3000 feet altitude would certainly give you a thought or two as you wrestle with the non existent controls on the way down..
How could we know well intentioned family planning services and welfare programs would in the long term destroy families and the earning power of the minority community?
I'm old, and I live in a rural area. I would love to see driving automated before I get too old to drive myself. I see Waymo cars on the road whenever I visit Phoenix, where they are being tested. In tech-friendly Arizona, they are just part of the scene.
If liberals would just eliminate the thugs and crazies on public transit, automated taxis could become the "last-mile' extension of transit, rather than being a rival to it.
I could also imagine a use-case for it with dedicated trucking lanes, though it seems to me trains would be cheaper.. But if trucks are running in their own area without passengers, then fine, a hack will only cost Wal-Mart a shipment of plastic junk or whatever.
Good points. "Autonomous driving" can be considered in many contexts and I'm mostly talking about an urban context given that this is where most of the people live and fatalities happen. Just from a purely technological view, I'd imagine the autonomous vehicles are already working well on wide open well lit low traffic AZ roads, and I'm certainly sympathetic to the needs of elderly/disabled folks. If anything this seems like the ideal use-case. That said, it doesn't necessarily assuage my concerns about big tech, hacking, etc. But, at the same time, if you have consented to the risks as an individual and want to get in the car on the road all by yourself, you aren't harming anyone. My concern would be, for example, a future where Los Angeles has a significant number of self-driving cars on the freeway all going at speed, a terrorist hack occurs, and thousands of people are killed. There are many people that state killing Americans as their lifetime goal and, normally, they'd have to actually get into the country to give it a whirl. I'm not aware of any commercial computer system that has not been hacked (banking, credit cards, even politician's emails), but bad PR or identity theft are limited to being annoying, not dismembering.
"Liability"..how does the issue with self-driving cars differ from the case of airplane operating under autopilot control? If an airliner is performing an autoland approach in bad weather, and the system fails to flare properly and runs it into the runway, who will be responsible for the damages?
Assuming the pilots set up the autopilot for the approach properly, then a court would probably divide responsibility among:
--the autopilot manufacturer
--the airframe manufacturer and the airline, depending on which of them selected that particular autopilot brand and model and how much due diligence they did in the selection
Complex systems do sometimes fail with bad results. This isn't a new problem.
The difference is scale. Passenger jets cost 80-200M USD, the value of the lives onboard is a secondary concern on a spreadsheet.
Right now you - and your insurance provider- could sue Google if one of it's car's kills you. How long you figure that'll last? Until they use their influence to write the Acceptable Casualty Law of 2028, capping damages as earning potential x life expectancy x 2?
The rates of accidents will decrease. That will be used to justify a host of profitable abuses of freedom and personal wealth unless we set the guardrails now.
Boeing paid 2.5 B to the US Government because they were so sloppy in covering up obvious fraud it was a layup for the DOT-OIG and FBI. 2 highly public, stock market cratering events. So your example is that it's okay, we should trust them becuase they killed people via fraud and got caught - and they're very sorry.
You are correct, crashing a plane is terribly difficult, you have like 5 miles to figure it out before you hit the ground. The monekys made great astronauts.
Ever been in a car crash in this country? Good luck getting a cop on scene under an hour. Think the Chicago field office and DOT inspector general is gonna spool up the choppers and send a special unit to pour over the black box when AI kills your wife in an self-driving car crash? To investigate Google or whatever tech giant and then actually go to bat against the world's best legal team? Nah bruh, you'll be a statistic, and they'll pay you what they think is fair. If you win some slanted appeals process.
I like a good counterexample as a challenge. I'm totally ignorant of airline technology but I'll say a few things that I don't think require expertise and if you're inspired you can tell me what you think. Using autopilot in planes must be different because:
- There are at least two pilots in the cockpit (on every plane I've ever been on) so they can take over if there's an issue. Taking over hundreds of thousands of cars that all go haywire all at once would require hundreds of thousands of employees to be on standby at all times. That won't happen, so cars will be on full autopilot.
- These systems have never been hacked for some reason and I can only imagine many terrorists would if they could. I'm very curious why this is but there are likely differences between this and AI cars, which I believe will run on consumer cellular networks. Perhaps airplanes have the hardware on board so they can't be hacked?
- Planes are just lower risk. They fly in 3D space without obstructions or millisecond surprises (kids, dogs, drunk drivers) on pre-plotted flight paths. They land on huge open runways limiting any landing failure to the plane itself. The FAA (or whoever does this stuff) grounds the whole fleet if there's a suspected issue.
- If all the cars in L.A. are hacked, there may be a few hundred thousands cars going full speed on the freeways all at once, or even, an entire country of cars on freeways.
So I'd say full self-driving rolled out at scale is a different animal of complexity, I'm not sure I would define a plane's autopilot as a complex system in that way.
There are indeed airplanes with autopilots with a single pilot...small jets, piston-engine planes, and some commuter liners. I don't know of any autopilot that requires external communication to function properly, and I'd hope such an autopilot wouldn't be certified. There IS a high and increasing % of aircraft navigation...including instrument landings..being done using GPS, which is fairly simple to jam. Pilots are supposed to also know how to navigate using the ground-based electronic beacon system and also by pilotage (looking out the window) in good weather.
Self-driving cars should not depend on cellular or any other network for their real-time operations. They will need connectivity for updates, and these could represent a hacking danger, possibly with a delayed-action virus to activate at a synchonized time.
My main point in my original comment is simply that assessing liability for automated systems isn't a new problem. In automotive, the failure of a software-based digital brake control system doesn't really seem any more difficult than assessing liability for an earlier analog-electronics based system.
Thanks for sharing as I don't know much about aviation. I don't know a ton about the autonomous car technology either but I just assumed that because companies like Waymo or Cruise can instantly commandeer a car remotely that they must be in constant communication using cell-phone towers or satellites or something, which seems like a security risk to me, as would the remote software updates.
That's a good point about the liability issue and it sounds right. Perhaps the point I was muddling towards is I do wonder if machine-learning systems are different, like for example if there is effectively no author to any given line of code as a result of machine-learning. And then in that case, how incentivized are individuals within that system to avoid risk? Who is held accountable for a deadly self-generated line of code, or one buried inside of infinite complexity by a bad actor? I don't know. It just seems like a different order of magnitude. As far as I know, there is nothing in my life at present that is both (a) run entirely by a large opaque software system that is also (b) capable of killing me when there's a software bug.
I am also an urbanist who wants to see car-centric landscapes designed to support more walking, biking, and transit. I don't see this agenda as threatened by replacing human drivers with robot drivers. Yes we need to accumulate data and report actual results (I posted above on how specifically to do this) but counting the sheer number of accidents due to specifically human errors such as sleepiness, substance abuse, and reckless non compliance with speed limits *strongly* suggests that errors specific to robot drivers will not get anywhere CLOSE to this tally. But yes we need to collect the data... while not pre-judging that robot driving is sabotaging urbanism.
The time to pass laws to compel Alphabet to post real data is before they establish market share worth several trillion (additional) dollars and a vested interest in protecting a semi-monopoly on all American transit. Google's lifeblood is data. That's what they are. Good luck getting it after the switch happens.
I wish them well, it's a worthy project and I agree that nobody wants to spend time in a car (driving or riding) vs living in an awesome safe walkable city. Lived in a couple, it's sweet.
However the past shows that the problems of society find a way to the city center, and anyone with a dollar finds their way out unless the government a) solves all of humanity's problems (Hungray repression time) or b) hires people with sticks to haul away the bums away to some kind of county farm (look for a Whole Foods, you can't miss it).
Fair points, my criticism in these debates is that we aren't talking about the big picture. I'll offer one example. Of course there's a gazillion I could pick but I'll choose my own street, where the speed limit has been reduced to 20mph as part of the "Vision Zero" plan, because road fatalities drop to near zero at that speed. Costs: However much street signs cost to replace, which is already done, and occasional low-speed injuries due to distracted driving. Solves problem of road fatalities that is the justification for the argument in the article above. Option 2: Replace all cars with AI. Cost of AI: Hundreds of billions of dollars meaning increases in car prices for the consumer, potential hacks, government ideological interference, handing over personal autonomy to companies that have proven over and over again they will abuse their power and collude with the federal government. To me it isn't at all obvious that autonomous driving is the right solution in the balance. Rather, I'd say that we live in a strange time when it's easier to imagine robo-taxis than having competent bureaucrats in city governance.
You raise some good points. In particular I think we probably need a constitutional amendment guaranteeing people the right to drive themselves if they want.
I think self driving cars will eventually be a great thing. Good for commuting, and especially good for elderly, or impaired drivers. But I agree about the government bad actor problem.
It’s fine that the author enjoys chatting on the phone or “unplugging” while being whisked around in a sterile, silent vehicle. I’d rather risk death by having a conversation with the human being who is driving me around. I’ve had some great talks, met some interesting folks, and gotten perspectives that are different from my own.
All this “if it saves one life” business is getting really old.
Eric is not considering the numbers correctly. Unless you compare the safety of self-driving cars on per car basis, you can't be sure which is safer. Thousands of people die in car accidents because there are millions of cars on the road. People have also died in self-driving car accidents even though there are a tiny number of these cars on the road. We need data to come to a reasoned conclusion.
I've always thought that the best use of this technology would be as a back-up system for faulty human drivers. Human distraction is probably one of the chief causes of accidents.
Which is why the new technology on vehicles like alarms that sound when you get to close to the dividing lines, alarms when you start to change lanes without signalling, etc. I've even heard of the vehicle actually pulling the wheel in the other direction!!! Yikes!
I find all of this to be annoying, but then, I am a capable and alert driver. I guess it would help with distracted drivers.
Once self driving cars are fully rolled out, their death total will probably be an order of magnitude less than human drivers. Computers can react quicker, plus they can communicate with all the other cars.
Note that's a different question than are self driving cars ready for a full roll out TODAY. The evidence for that is less clear.
"We are all bad drivers? Speak for yourself. I am in many ways glad that I will not likely live long enough to spend my time on the road dodging computer controlled vehicles. Even my adaptive cruise often drives me batty!
“ Anyone rooting against self-driving cars is cheering for tens of thousands of deaths, year after year.” If you disagree with me, you’re evil. Garbage.
In mentioning the thousands of accidents, Mr. Newcomer does not distinguish between those which happen in low speed areas versus on the highway. Robots are good highway drivers in terms of safety but lousy highway drivers from the perspective of human drivers.
My 2022 car when operating in "Adaptive Cruise Control" mode (as a simple robot) on the highway responds as it should for safety to a slowing car ahead, by sacrificing speed for distance to maintain adequate calculated braking distance. This "correct" action is the opposite of what humans do, who generally sacrifice distance to reduce speed as little as possible. As such my good robot driver slowing down more than a human creates an unexpected and potential dangerous situation behind me and at a minimum annoys the driver following me.
What happens next as traffic speeds up is my robot does not speed up as quickly to avoid aggressive acceleration and so to the driver behind me I am annoyingly not keeping pace.
Note that in both cases, the robot is driving correctly and safely. Should the programmers teach it to drive like a human or someone's grandmother and stay in the right hand lane? Would you like to be in that car?
Worse, when the highway curves to the left while an exit is straight ahead, my car sometimes misinterprets the car in front of me slowing down to exit as an abruptly slowing car on the highway. In response, my car breaks heavily and unexpectedly, startling the driver behind me who is hopefully paying attention and who sees no reason or warning for me to brake. Rage? I have nearly been re-ended twice in this scenario.
The more robots, the more grandmothers there are on the highway which I submit is not a better safety environment. Perhaps a bumper sticker would be appropriate. "Don't Blame Me, My Robot is Driving"
I think the bigger thing is that human drivers naturally tend to drift to the other side of the lane of a car when passing to get a bit more distance. Robots don't do that. Which is probably not the right course of action when other human drivers are still on the road.
Self driving vehicles are inherently privacy invasive. That ought to give pause of itself.
Also, when a driver causes a homicide we understand that it's a failing of that individual. However, when an autonomous vehicle causes a fatality it's because of what a programmer or engineer thought was an appropriate trade off when training the engine.
They are different categories of both culpability and morality.
Not absolutely against, but it's not just a straight utilitarian issue.
It's worse than that though because any given failure may not be traceable to a specific decision made by a specific set of people. The systems are black boxes by design - a giant pile of training data is ingested, resulting in a bunch of inscrutable neural net weights. It's usually going to be impossible to trace some action of one of these systems back to some particular bit of training data. Why did ChatGPT say that totally made up thing? No one knows, and it certainly wasn't in the training data.
Surely Alphabet, Inc. will be transparent with the data - and inherent liability it corroberates and will not use it's golliath power to lobby for laws that prevents normal people harmed by this technology (an acceptable sacrifice, on the net! will save thousands of lives, no argument!) from seeking recourse that would otherwise bankrupt it durring the adoption phase
How sad to see such a technologically illiterate article in a magnificent journal like The Free Press. The Free Press has an admirable moral compass, but it is weak in the realm of real world technology. How absurd is the proposition that “X percent (pick a number) of traffic accidents are caused by human drivers. Therefore eliminating human drivers will reduce traffic accidents”? The pursuit of self-driving technology has yielded some great driver-assist systems, but the claim of self-driving, except under the most constrained conditions, will yield only carnage. Free Press, you can do better than this.
In the late 'twenties of the last century, a barnstormer crossed the Midwest, landing at farms along the way and giving rides to locals. An old farmer, driving his horse and flatbed wagon, brought his young son to see the spectacle.
He approached the pilot rather skeptically and eyed the machine up and down. Finally he rubbed his chin and said, "What good is it?"
The barnstormer pointed to the young son standing by the old man's side and said, "What good is HE?"
"Why, he's just a child now, but someday he will be a man!"
Author: "Anyone rooting against self-driving cars is cheering for tens of thousands of deaths, year after year."
Really... you are "cheering for death" if you don't agree with the move toward less individual freedom and more government/corporate control of human life? This kind of hyper-emotional attempt to shame those who hold another opinion tells me that I can disregard the rest of this guy's argument.
I'm not a luddite. But the safety justification better not open the door to taking away my autonomy. I like driving. Take a lesson from unusable gas containers, low pressure toilets and showers.,
Let's also take a lesson from the humble traffic circle. Reduces accidents by 66%, fatalites by like 92%, emissions by like 40% because your car is least efficient from 0-10mph and you can kinda keep rolling through it. And pedestrian fatalites. And stress. And you can use 1 lane instead of 2-3 at the same efficiency.
If this was purely data-driven, THINK OF ALL THE LIVES THIS WILL SAVE we'd tear down every stop sign and traffic light in the country and have nothing but traffic circles because we can do that in 6 months instead of waiting .... how many years now? For waymo to get thier shit on the road and save us from ourselves.
You know it's going to happen. You're a bad driver, don't you see? You shouldn't be allowed on the road like an animal. Or you're literally murdering people.
"The innocent self-driving car was burnt to a crisp."
Tip #1 for reaching people on the other side of this issue: do not anthropomorphize or ascribe moral value to technology. It smells weird and gets the hackles up. Reserve moral categories for the humans creating and using the technology.
A normal car is a physical extension of it's owner that Americans feel a deep emotional connection to, so I could see how the author could fall into this mistake.
A Waymo car is the physical extension of Alphabet, Inc. - and Google has in 70 billion ways not covered itself in glory this month on the "trust us" front.
Honestly I wish more tech companies would offer us physical manifestations that we can publicly immolate.
“ Anyone rooting against self-driving cars is cheering for tens of thousands of deaths, year after year.”
I guess I'm cheering for tens of thousands of deaths, year after year, then.
Utilitarianism is evil. Automating everything and delegating to AI takes away human agency. You'll have to rip my steering wheel from my cold dead hands.
"And contrary to the view that driverless cars are being deployed unilaterally by tech billionaires, the people’s representatives—government officials—gave Alphabet-owned Waymo a license to operate."
Typical leftist Utopian soft totalitarian thought process.
People just shouldn’t be allowed to do things the Oligarchy deems dangerous. It’s for people’s own good. It’s for the innocent children. It saves money. Exceptions for Nomenklatura of course.
1. Luxury belief. The people advocating most loudly for implementing these are the ones that do not live in the places these are tested. You’re welcome to donate your neighborhood and children to the testing curve that goes with implementation of any new technology. After all if your child dies you can console yourself that, statistically, it’s for the greater good.
2. Lack of data. These have never been deployed at scale. Saying they are safer than human drivers (implication: at scale) is like saying a barely tested vaccine is safe and effective. Maybe, maybe not. This claim is just scientifically invalid.
3. Liability matters. Right now at least there is the threat of prosecution for dangerous driving. Which court, exactly, is equipped to comb through billions of lines of code to discern who messed up? As far as I know there’s never been an autonomous system of any kind holding millions of lives all at once in any given moment, and what the incentive structures look like when liability is decoupled from individual actors. And are you sure that amongst tens of thousands of engineers there will never be one horrifically stupid or bad actor?
4. Techno utopians remain shockingly ignorant that technologies never develop as planned. Look at the horror that is too often Facebook. Or smartphones on kids. Technologies’ second, third, fourth order etc effects are inherently unpredictable. The creators are never held accountable.
5. Individuals have autonomy. In contrast, a massive system is different and susceptible to hackers beyond individual control. Generally people don’t get dismembered when their bank account is hacked. How about a hack with hundreds of thousands of drivers all doing 65 mph?
6. I’ve said the solution to this for years is not to make Americans more sedentary by car expansion but restructure cities to be human and child centric. Most Americans live in cities. Many cities in the world have solved this problem already with great urban planning and public transport. Cars are a bad idea for a lot more than road fatalities, in many ways they’ve diminished the beauty and livability of American cities, health, and human relationships.
7. Bad government actors. We’ve seen totalitarian overreach in many areas and it’s even worse in Canada. If governments control social media accounts and cancel bank accounts for political persecution, what exactly does a protest look like when those same corrupt government officials are installed at Waymo just like they were at Twitter?
My point: human freedom isn’t possible when large opaque systems can in one moment derail someone’s intentions or life. There are many other highly successful interventions already in use to reduce traffic deaths that don’t have any of the above risks. Even with a car centric city like Portland they’ve reduced many of the speed limits to 20mph, for example. AI driving is a different animal from past technologies in many dimensions and should therefore be in its own category and not naively equated to the rollouts of past technologies.
Re: (7) .... imagine the future in which all travel is by automated car and no personal vehicle ownership is permitted (the future they want). They can make you a virtual prisoner by shutting off your vehicle access. They can prevent you from traveling certain places at certain times, or if an event is happening.
The fact that Waymo is owned by the same company as Google gave me pause.
Waymo:
“I’m sorry, I am unable to offer you my transportation services because you do not fit into our goal of a diverse passenger infrastructure.”
😂
You: "Take me to Chik-fil-A."
Your Waymo car: "I'm sorry, Dave . . ."
In addition, Waymo is in a strategic partnership with Zhejiang Geely Holding Group in China.
We will all live like serfs eating bugs in futuristic, high-tech, "15 minute cities".
Some of us will be outcasts in the wilderness. I’m good with that.
Your point 6 sounds a lot like the 15-minute cities that Globalists want to cram us into.
Part of the reason that public transit works better in Europe is because the much denser population can sustain it. That is not true of the vast majority of American cities. I've tried to get around mid-sized American cities with public transit, and it's a much different (and far more inconvenient) experience than my experience with public transit in European cities.
The reality is that most American cities lack the population density to make public transit routes that go where a lot of people need to go practicable. One *could* design the necessary routes and scheduling, but many of the buses would be traveling empty or nearly so too often to justify the expense involved.
I would dearly love to live in a 15-minute city, which was effectively made illegal in most U.S. jurisdictions for the past six or seven decades.
I've gotten as close as I can to it: I can commute to work by train, walking to and from the station on both ends. The one thing I can't do on foot is to shop for things I need, not even groceries. I mean, there are lots of twee boutiques in town, but man cannot live by sustainably-sourced hand-knit wool sweaters alone.
When I lived in France, in the northern suburbs of Paris, in the early 1990s, one of the very best things about it was that there was a bakery and a small grocery, side by side, directly across the street from my apartment . . . and a commuter rail station just down the street.
That was a good and decent way for a human being to live, and it didn't even have to be forced by (allegedly) tyrannical planners. All you have to do is not make that stuff illegal, not force zoning and planning codes that separate different land uses from one another, move things farther apart, and push them farther back from the street. In fact, the most discordant and dysfunctional parts of the northern suburbs of Paris are around the big ugly tower blocks in the style of le Corbusier, who wanted to impose his own modernist, car-centered vision on everyone.
What I'm describing are just the natural forms human settlements took in the past, before the regimentation of zoning and the effects of the automobile.
Agreed mixed use zoning can be a great thing.
Most Americans live in suburbs(69%) and only 12% in urban areas.
I live in a suburban city and city buses usually are about 5% full all the while spewing diesel exhaust into the air. The city's infrastructure wasn't designed for buses so when a bust stops at a bus stop half of the bus is sticking out into the lane which backs up traffic and causes safety issues as drivers try and swerve around them.
The last time I was in SF I rode in a driverless car and it was nice. The car was clean, got to the destination in the most efficient route and was safe.
You're totally right about the density issue and I wouldn't want the 15-minute city type of central planning either, because personally I'm against any mass-scale bureaucratic utopia. But I love this topic so let me add some context to my above comments.
In effect, the present world is the fruition of a past utopia, the one envisioned by previous bureaucrats and heavily influenced by the auto lobby. The freeways carved through American cities destroyed majority black neighborhoods (and also European Ethnic neighborhoods like Greeks and Italians in Chicago, for example), not to mention many waterfronts. Definitely wasn't democratic. If I recall my history, GM deliberately dismantled LA's formerly great public transport system (and faced anti-trust lawsuits) and in New York Robert Moses built roads specifically to contain minorities in urban areas, among many examples. The car-centric city was foisted upon Americans through federal and local (zoning) policies, motivated by a variety of ideologies, and certainly not crowd sourced in a free market. Yet prior to this policy shift, urban mass transport networks (often private) thrived, decades after car ownership was widespread. I'm not an expert in these areas so I can't weigh all the factors, and certainly voluntary car adoption was one factor of how we got here, but, my point is that American cities haven't been allowed to grow organically in the first place.
Looking at the details, the reason many American cities lack density is because of explicit decades-long zoning laws banning it. This has recently changed in some places like Portland but the barrier remains still that, for example, a "cheap" $350k 1-bedroom in SF costs exactly twice that after all the bureaucrats get done charging bogus permitting fees. It can take years to develop a parcel as a result. Yet Americans want density. Some of the most expensive neighborhoods in the country are the densest. People voluntarily move to these neighborhoods and pay huge percentages of their incomes to live there. People want to live in fun places where their kids can walk around without cars driving fast. People want to be able to walk to the grocery store, not because of eco-ideology but because it's really enjoyable. Americans visit Europe by the millions just to experience this sort of thing. So in my view things like density are downstream from a policy problem which creates a density problem, and the density problem creates high speed transport infrastructure leading to an excess road fatality problem. The criticism I have of the above articles is that it just quotes one statistic (road fatalities), out of context, and then uses that to justify a large-scale transformation of our entire society. We see this pattern of argument in many areas but I'll leave that for another time. I'd like to live in a world that is neither an auto-corporation utopia nor a globalist utopia, and certainly not an AI utopia to compensate for the failures of the auto utopia. So this is what I mean by the solutions being so much simpler than all that: let Americans build out their own cities in relative freedom. Density will take time to build out but we need to take the long view.
I disagree with your assertions about car-centric cities not being "crowd sourced in the free market." While specific details of highway planning are due to bureaucrats, the demand for those highways came from car owners.
The automobile made it possible for people to live further from their workplace. It led directly to suburbs and urban sprawl. Although *you* may love cities, with all their crowding and noise (and now, increasingly, crime and filth), many people do not.
In the late 19th and early 20th century, there was a mass of population movement from the country to the city. Young people were sick of the hard work involved with farming, and cities offered many things that the country could not. But as cars were becoming more and more available, many of those people realized that high population density was not so pleasant to live in night and day. They yearned to have a bit of the country they'd grown up in, and cars made it possible to live in the country areas near the city. Housing developers met the rising demand. Suburbs were born.
High demand for expensive housing in dense urban areas is driven by people who thrive on city life, who don't mind living in apartment buildings with many (not always polite) neighbors, and who don't mind crowding. In some places (like SF) there is status associated with such neighborhoods.
But a lot of people *hate* living like that. You literally could not *pay* me in live in such conditions.
It sounds like you are angry about cities. Honestly, it would be nice for you to have thoroughly read my post before commenting as you are responding to points that I didn't make, and missing points that I did make, including regarding the history of suburbanization and individual choice. But in general I'm not interested in debating whether or not you enjoy living in cities. My point is to challenge the author's simple-minded assertion that the solution to traffic deaths is AI when there are many policy decisions that contribute to traffic deaths, particularly in sprawling urban areas where the majority of Americans live. Again, for example, if I could vote between (a) spending hundreds of billions on AI or (b) allowing individuals to build out urban density because they want to, I'd choose b.
I generally agree with your points here, except I think you're still giving too much credit and leeway to those "techno utopians" you deride. Walkable cities are fine if that's what you prefer, but not everyone wants to live in a city (some of us prefer quiet and wide-open space, farmable land, and polite people). The automobile is one of the greatest markers of freedom and autonomy, especially in a country as big and varied as the US. That's one of the reasons widespread public transport never took off. If you wake up one day and want to go to the beach (but live in OK or CO), you can start driving without having to check train schedules or anything.
As for driverless cars, getting in one means you are immediately at the mercy of a machine controlled by someone you don't know and can't see. It's nightmarish. Not only has the tech company run your background when you paid for the car service, now they have you in a speeding death trap that can take you anywhere they want. They can drive you into the ocean, into a brick wall, into oncoming traffic. At least with a human driver you have the option to try and stop them. Or, you know what, why not try this great thing - don't know if you've heard of it - but drive the car yourself. Then you can make your own decision to drive into the ocean if you'd like.
If you're a White person being driven by a DEI AI, you're in big trouble.
Those will automatically lock the doors and drive you to the re-indoctrination center. If you are Jewish, it will fill with the output of a small internal combustion engine.
The whole "freedom" argument for driving is very weak when one considers that driving requires vast subsidies and infrastructure and has enormous effects on those who do not drive as well as those who do. Your argument about "immediately being at the mercy of a machine controlled by someone you don't know" can be applied to airlines as well and we know they are the safest form of travel. I don't like arguments that put up a wall to protect the status quo. You cannot use a smarmy "freedom" argument to imply that there is unmanaged danger in robot vehicles while making no attempt to compare it to the unmanaged danger of the status quo. There is NOTHING that protects you and me from the driving of people who are drowsy, inebriated, high, suicidal, or just plain reckless. We are powerless against this "freedom".
As a conservative, protecting the status quo - when worth protecting - is the point. I look at the alternatives, weigh the pros and cons, and make a decision if the thing we're doing now is the best among other options. Progressives (which includes tech utopians), on the other hand, barrel through without considering the consequences of what they're destroying. The automobile was revolutionary in a way progressives can only dream of. But what it was replacing (slow/inefficient/relatively expensive forms of travel) was vastly inferior. Over time, new inventions and regulations made them more safe. And the great thing about that is, if you are so inclined, you're not forced to drive the safest car. There was an article in WSJ the other day about a guy who restored a Model A and drives it whenever he wants, even though it has little to no of the features of a modern car (he compared it to driving a tractor). And he has the freedom to do so.
I'm not even arguing that "autonomous" cars won't eventually be as safe or safer than cars are now (one of the commenters here mentioned that because there are so little data, any safety claims about them are impossible to prove). But, in the here and now, not some dreamed of utopia, and as long as I have the freedom to do so, I won't be subjecting myself to the whims of a new technology (compared to planes) that I have little to no control over.
Yes, I recognize that vast sums of federal and state dollars goes to maintaining the interstate and roads and gas stations/oil industry, but it is in the service of allowing myself and other citizens the freedom to drive wherever and whenever we like. To move from one state to another. To not be locked in our homes subsisting on virtual reality and food delivery. And yes that freedom comes with risks from stupid and reckless people, but I still prefer it to the alternative. You have the freedom to disagree.
I certainly can see why some are "late adopters" or even "never adopters" to a new technology. I will choose to an early one and that makes it better for the later ones. The crux of the disagreement is that those of us who favor robot driving will want the system to cater to our needs versus the needs of human drivers. Yes real data is necessary to prove a safety case but we must allow significant deployment of robot driving for such real data to be collected.
Does choosing "smarmy" as the adjective for your scare-quoted noun "freedom" belie a desire to have Google and/or the government, take your smarmy freedoms away in return for your supposed safety?
You are not powerless against people who are drowsy, inebriated, etc. Shame on whoever told you you are powerless. Highly skilled and situationally aware drivers are extremely good at avoiding collisions with such drivers. Not with 100% effectiveness, but a self-driving car isn't offering that either. And self-driving cars haven't been shown to be good at protecting you from any of those things. Unless, of course, you posit that they replace all human drivers.
Human agency, properly deployed, is vastly safer than anything anyone in the self-driving car industry even feigns will be on offer any time soon.
It amazes me how willing some people are to hand over to robots some of the very things that makes us human: our autonomy and our agency.
It is not a surprise that the people who are anxious for us to give our agency and autonomy to the likes of Google have been all too excited to give it to our government (see the Covid response, for example). It is not a coincidence that almost all Googlers, and government employees, are of the Democratic party. They believe Big Brother should be watching you, controlling you, thinking for you. And, of course, they imagine themselves so brilliant they should be Big Brother.
Freedom is a gift from God - you have a natural right to it in every circumstance, not because some government or company gave it to you. You don't need to trade it for security.
We are also powerless against Alphabet, Inc. which routinely abuses it's power over every freedom it is given accesss to and is the frontrunner in driverless car technology.
What does protect me from the other drivers on the road currently - however poorly - is a shitload of good laws concerning personal liability, freedom of movement, a training requirement, an insurance requirment and armed state officials enforcing these rules 24/7 with (semi) strict oversight themselves. Nowhere else in my private life am I subject to such intense scrutiny.
When I get into a Google car, I am abandoning all legal protection to their terms of service unless we put up a similar legal framework. Same as an airline, as you say.
I want driverless cars as bad as you do. I want to go places while I'm drinking without putting someone else in danger. I want my older family members to have freedom of movement. I have been injured in car wrecks, my family as well, and I've had freinds and neighbors killed.
But in the same way you wouldn't have the Sackler family that made the heroin crisis solve the pandemic (fuck, wait we did that), or the CIA determine our freedoms with the patriot act (fuck, again?) don't let Google play us for fools using fear to sell a product that needs strict oversight to prevent abuse.
With all of those safeguards you mention, there are about 40K deaths per year. I don't know anyone in my life who has been murdered, Zilch. I have had a conservation with close to a dozen people killed by a car. Is this experience any different from you? Humans driving a machine weighing over a ton at any speed they chose on their whim is literally one of the most dangerous activities that we commonly do. If I made a bet about lives killed per mile driven by human vs mile driven by robot, would you take the human side of the bet? I am so confident of the result, I would give you at least 3x odds. :-)
Enjoy living in your bubble wrapped safe world free from “harms”.
“You cannot use a smarmy ‘freedom’ argument to imply that there is unmanaged danger in robot vehicles while making no attempt to compare it to the unmanaged danger of the status quo.” Sir, I will have you know that this is The Free Press comment section where freedom arguments are our bread and butter. LOL
I think we agree. I'm not universally anti-car and of course there are times when cars are necessary, certainly for rural areas. Personally I love a good road trip. There are people that aren't able-bodied, etc, and they need help. Rather my points are challenging the idea that AI is the best solution to the road fatality problem. I wrote an additional comment above about that and urban density.
Funny you should mention this. Just saw that a Hawaii tourist driving a rental car and following his GPS drove over a cliff and into the ocean requiring rescue from the coastguard. Delivery trucks following GPS regularly drive up my driveway which is on the roadbed of a county highway discontinued over 50 years ago but apparently still shown on old USGS maps. Not putting my trust in GPS navigation.
The article never addresses the 226 accidents caused by driverless cars in San Francisco. Horrific accidents have involved robocars hit and runs and a pedestrian trapped under an unmoving robocar.
The author also claims that remote corrections happen efficiently, this is not always true. Ambulances have been blocked behind robocars and the robocars have mysteriously stopped causing huge traffic delays.
The anger against robocars is not a situation of angry peasants suspicious of new technology, but flaws in robocars causing accidents, blocking bike lanes, and behaving badly and dangerously. The author’s failure to address the flaws makes this a pr piece not a report.
"Accidents" are a far cry from fatalities. Car accidents are so common that no one bothers to count them. We don't even need a police report for an accident unless there is a human injury. The context here is serious human injury, not fender benders.
The NHTSA reported 392 crashes in a 10-month period involving cars with self-driving technology. Of these, there were six fatalities and five serious injuries.Sep 30, 2023 Granted that this stat may not separate self driving from cars without drivers, but there are problems.
A woman was ran over by a self driving car and it pinned her to the pavement. Emergency crews had to remove her. The self driving cars have blocked ambulances and ran over fire hoses at emergency scenes. The self driving cars have blocked traffic including bus and bike lanes for hours. While the author experienced one easy correction, his single experience does not reflect what always happens.
If you are going to write about technology, then address all the issues including its impact on SF traffic and not a puff piece that reads like pr rather than a report.
This is a hilariously timed article.
AI is probably the future of humanity, and a trillion dollar arms race favors the big dogs in the fight. It's also essential to the operation of self driving cars, which Google leads in.
Google lost $70 BILLION in market cap this week because it's AI launch couldn't make pictures of white people and called Norman Rockwell a white supremacist. Had to pull the plug after a day.
In $ value and prestige lost in 24 hours, that's the equivalent of the US Navy launching SIX new USS Ford class aircraft carriers on one day and having them all sink immediately because it would be racist to close the hatches or check the welds.... wait does Boeing make boats?
At this point I'm not sure our prospective evil tech overloards have the technical comptence to actually make the pods or fake burger bugmeat to force feed me.
Actually most Americans live in the suburbs and public transit is of little or no benefit to them.
I’m in favor of self driving cars and of course it’s idiotic for people to destroy them. I think it would be something of a utopia to have all of the freedom of a car without having to actually drive it.
However, as to your point 2 - right now, statistically, self driving cars are more dangerous than human driven cars on a miles driven basis.
There is 1 fatality per every ~90 million miles driven in the US. Self driving cars haven’t yet driven 90 million miles and have killed one person.
Now it’s true that the car that killed a person wasn’t a Waymo, and it’s also true that it happened several years ago.
But it’s also true that self driving cars only operate in a handful of urban locations (maybe only SF and Chandler AZ?), and not in places where ice, snow, or narrow rural roads are an issue.
Again, none of this is to denigrate self driving cars - only the ridiculous assertion that they are, right now, safer than human driven cars. They may be, but we just don’t have enough evidence yet.
I love the clarity of your comment!
All worthy points, Richard. Especially #5. I would be curious how enthusiastic our intrepid author would be sitting at the front passenger seat of a self driving commuter airplane, coming soon I hear. A hack at 3000 feet altitude would certainly give you a thought or two as you wrestle with the non existent controls on the way down..
Haha, good point. Well with 3000 feet of altitude there'd be a good minute left for contemplating existential risk of AI technologies.
“Technologies’ second, third, fourth order etc effects are inherently unpredictable. The creators are never held accountable.”
Non sequitur. if consequences are unknowable, how can creators be held accountable for them?
I mean it's usually pretty easy to guess
"How could we know prescribing legal heroin en masse would lead to addiction?"
"How could we know [insert social media platform] would be a haven for pedos?"
When it's physical harm, criminal charges are possible.
Digital harm? It's in the terms of service my dude.
:How could we know open borders would lead to astronomical drug and human smuggling, and the disruption of social resources?"
How could we know well intentioned family planning services and welfare programs would in the long term destroy families and the earning power of the minority community?
We could go on.... and on.... and on....
The road to hell does go on and on.
I'm old, and I live in a rural area. I would love to see driving automated before I get too old to drive myself. I see Waymo cars on the road whenever I visit Phoenix, where they are being tested. In tech-friendly Arizona, they are just part of the scene.
If liberals would just eliminate the thugs and crazies on public transit, automated taxis could become the "last-mile' extension of transit, rather than being a rival to it.
I could also imagine a use-case for it with dedicated trucking lanes, though it seems to me trains would be cheaper.. But if trucks are running in their own area without passengers, then fine, a hack will only cost Wal-Mart a shipment of plastic junk or whatever.
Good points. "Autonomous driving" can be considered in many contexts and I'm mostly talking about an urban context given that this is where most of the people live and fatalities happen. Just from a purely technological view, I'd imagine the autonomous vehicles are already working well on wide open well lit low traffic AZ roads, and I'm certainly sympathetic to the needs of elderly/disabled folks. If anything this seems like the ideal use-case. That said, it doesn't necessarily assuage my concerns about big tech, hacking, etc. But, at the same time, if you have consented to the risks as an individual and want to get in the car on the road all by yourself, you aren't harming anyone. My concern would be, for example, a future where Los Angeles has a significant number of self-driving cars on the freeway all going at speed, a terrorist hack occurs, and thousands of people are killed. There are many people that state killing Americans as their lifetime goal and, normally, they'd have to actually get into the country to give it a whirl. I'm not aware of any commercial computer system that has not been hacked (banking, credit cards, even politician's emails), but bad PR or identity theft are limited to being annoying, not dismembering.
"Liability"..how does the issue with self-driving cars differ from the case of airplane operating under autopilot control? If an airliner is performing an autoland approach in bad weather, and the system fails to flare properly and runs it into the runway, who will be responsible for the damages?
Assuming the pilots set up the autopilot for the approach properly, then a court would probably divide responsibility among:
--the autopilot manufacturer
--the airframe manufacturer and the airline, depending on which of them selected that particular autopilot brand and model and how much due diligence they did in the selection
Complex systems do sometimes fail with bad results. This isn't a new problem.
The difference is scale. Passenger jets cost 80-200M USD, the value of the lives onboard is a secondary concern on a spreadsheet.
Right now you - and your insurance provider- could sue Google if one of it's car's kills you. How long you figure that'll last? Until they use their influence to write the Acceptable Casualty Law of 2028, capping damages as earning potential x life expectancy x 2?
The rates of accidents will decrease. That will be used to justify a host of profitable abuses of freedom and personal wealth unless we set the guardrails now.
And in terms of consumer abuses I couldn't think of a worse industry to pick to defend carte blanc to the carrier than the friggin airlines
There is no safer way to travel, by a long shot.
Boeing settlement for the two 737 Max crashes was $2.5 billion.
Boeing paid 2.5 B to the US Government because they were so sloppy in covering up obvious fraud it was a layup for the DOT-OIG and FBI. 2 highly public, stock market cratering events. So your example is that it's okay, we should trust them becuase they killed people via fraud and got caught - and they're very sorry.
You are correct, crashing a plane is terribly difficult, you have like 5 miles to figure it out before you hit the ground. The monekys made great astronauts.
Ever been in a car crash in this country? Good luck getting a cop on scene under an hour. Think the Chicago field office and DOT inspector general is gonna spool up the choppers and send a special unit to pour over the black box when AI kills your wife in an self-driving car crash? To investigate Google or whatever tech giant and then actually go to bat against the world's best legal team? Nah bruh, you'll be a statistic, and they'll pay you what they think is fair. If you win some slanted appeals process.
I like a good counterexample as a challenge. I'm totally ignorant of airline technology but I'll say a few things that I don't think require expertise and if you're inspired you can tell me what you think. Using autopilot in planes must be different because:
- There are at least two pilots in the cockpit (on every plane I've ever been on) so they can take over if there's an issue. Taking over hundreds of thousands of cars that all go haywire all at once would require hundreds of thousands of employees to be on standby at all times. That won't happen, so cars will be on full autopilot.
- These systems have never been hacked for some reason and I can only imagine many terrorists would if they could. I'm very curious why this is but there are likely differences between this and AI cars, which I believe will run on consumer cellular networks. Perhaps airplanes have the hardware on board so they can't be hacked?
- Planes are just lower risk. They fly in 3D space without obstructions or millisecond surprises (kids, dogs, drunk drivers) on pre-plotted flight paths. They land on huge open runways limiting any landing failure to the plane itself. The FAA (or whoever does this stuff) grounds the whole fleet if there's a suspected issue.
- If all the cars in L.A. are hacked, there may be a few hundred thousands cars going full speed on the freeways all at once, or even, an entire country of cars on freeways.
So I'd say full self-driving rolled out at scale is a different animal of complexity, I'm not sure I would define a plane's autopilot as a complex system in that way.
There are indeed airplanes with autopilots with a single pilot...small jets, piston-engine planes, and some commuter liners. I don't know of any autopilot that requires external communication to function properly, and I'd hope such an autopilot wouldn't be certified. There IS a high and increasing % of aircraft navigation...including instrument landings..being done using GPS, which is fairly simple to jam. Pilots are supposed to also know how to navigate using the ground-based electronic beacon system and also by pilotage (looking out the window) in good weather.
Self-driving cars should not depend on cellular or any other network for their real-time operations. They will need connectivity for updates, and these could represent a hacking danger, possibly with a delayed-action virus to activate at a synchonized time.
My main point in my original comment is simply that assessing liability for automated systems isn't a new problem. In automotive, the failure of a software-based digital brake control system doesn't really seem any more difficult than assessing liability for an earlier analog-electronics based system.
Thanks for sharing as I don't know much about aviation. I don't know a ton about the autonomous car technology either but I just assumed that because companies like Waymo or Cruise can instantly commandeer a car remotely that they must be in constant communication using cell-phone towers or satellites or something, which seems like a security risk to me, as would the remote software updates.
That's a good point about the liability issue and it sounds right. Perhaps the point I was muddling towards is I do wonder if machine-learning systems are different, like for example if there is effectively no author to any given line of code as a result of machine-learning. And then in that case, how incentivized are individuals within that system to avoid risk? Who is held accountable for a deadly self-generated line of code, or one buried inside of infinite complexity by a bad actor? I don't know. It just seems like a different order of magnitude. As far as I know, there is nothing in my life at present that is both (a) run entirely by a large opaque software system that is also (b) capable of killing me when there's a software bug.
I am also an urbanist who wants to see car-centric landscapes designed to support more walking, biking, and transit. I don't see this agenda as threatened by replacing human drivers with robot drivers. Yes we need to accumulate data and report actual results (I posted above on how specifically to do this) but counting the sheer number of accidents due to specifically human errors such as sleepiness, substance abuse, and reckless non compliance with speed limits *strongly* suggests that errors specific to robot drivers will not get anywhere CLOSE to this tally. But yes we need to collect the data... while not pre-judging that robot driving is sabotaging urbanism.
The time to pass laws to compel Alphabet to post real data is before they establish market share worth several trillion (additional) dollars and a vested interest in protecting a semi-monopoly on all American transit. Google's lifeblood is data. That's what they are. Good luck getting it after the switch happens.
I wish them well, it's a worthy project and I agree that nobody wants to spend time in a car (driving or riding) vs living in an awesome safe walkable city. Lived in a couple, it's sweet.
However the past shows that the problems of society find a way to the city center, and anyone with a dollar finds their way out unless the government a) solves all of humanity's problems (Hungray repression time) or b) hires people with sticks to haul away the bums away to some kind of county farm (look for a Whole Foods, you can't miss it).
Fair points, my criticism in these debates is that we aren't talking about the big picture. I'll offer one example. Of course there's a gazillion I could pick but I'll choose my own street, where the speed limit has been reduced to 20mph as part of the "Vision Zero" plan, because road fatalities drop to near zero at that speed. Costs: However much street signs cost to replace, which is already done, and occasional low-speed injuries due to distracted driving. Solves problem of road fatalities that is the justification for the argument in the article above. Option 2: Replace all cars with AI. Cost of AI: Hundreds of billions of dollars meaning increases in car prices for the consumer, potential hacks, government ideological interference, handing over personal autonomy to companies that have proven over and over again they will abuse their power and collude with the federal government. To me it isn't at all obvious that autonomous driving is the right solution in the balance. Rather, I'd say that we live in a strange time when it's easier to imagine robo-taxis than having competent bureaucrats in city governance.
Slight correction to your last statement: I think it actually is easier to imagine competent robo-taxis than competent bureaucrats in city governance.
You raise some good points. In particular I think we probably need a constitutional amendment guaranteeing people the right to drive themselves if they want.
I think self driving cars will eventually be a great thing. Good for commuting, and especially good for elderly, or impaired drivers. But I agree about the government bad actor problem.
Thanks for your thoughtful review.
Trading freedom for safety, no thanks
It’s fine that the author enjoys chatting on the phone or “unplugging” while being whisked around in a sterile, silent vehicle. I’d rather risk death by having a conversation with the human being who is driving me around. I’ve had some great talks, met some interesting folks, and gotten perspectives that are different from my own.
All this “if it saves one life” business is getting really old.
"There’s a natural, though irrational, human bias toward the status quo. We tend to believe that things are the way they are for a good reason."
Good point!
Now, let's try your impeccable logic on the following things that should be dismantled, destroyed or just razed to the ground:
Public Schools
Teacher's Unions
The Democratic Party
All other labor unions
Chicago (the entire city)
As a long suffering local I request the current Carolina Panthers team and staff be humanely euthanized and the stadium destroyed
Eric is not considering the numbers correctly. Unless you compare the safety of self-driving cars on per car basis, you can't be sure which is safer. Thousands of people die in car accidents because there are millions of cars on the road. People have also died in self-driving car accidents even though there are a tiny number of these cars on the road. We need data to come to a reasoned conclusion.
I've always thought that the best use of this technology would be as a back-up system for faulty human drivers. Human distraction is probably one of the chief causes of accidents.
I agree: the data surely shows our cell usage has spiked accidents.
Which is why the new technology on vehicles like alarms that sound when you get to close to the dividing lines, alarms when you start to change lanes without signalling, etc. I've even heard of the vehicle actually pulling the wheel in the other direction!!! Yikes!
I find all of this to be annoying, but then, I am a capable and alert driver. I guess it would help with distracted drivers.
Hey, it worked for Michael Hastings! 😳
Once self driving cars are fully rolled out, their death total will probably be an order of magnitude less than human drivers. Computers can react quicker, plus they can communicate with all the other cars.
Note that's a different question than are self driving cars ready for a full roll out TODAY. The evidence for that is less clear.
"We are all bad drivers? Speak for yourself. I am in many ways glad that I will not likely live long enough to spend my time on the road dodging computer controlled vehicles. Even my adaptive cruise often drives me batty!
“ Anyone rooting against self-driving cars is cheering for tens of thousands of deaths, year after year.” If you disagree with me, you’re evil. Garbage.
Agreed. Strong indication of weak arguments.
In mentioning the thousands of accidents, Mr. Newcomer does not distinguish between those which happen in low speed areas versus on the highway. Robots are good highway drivers in terms of safety but lousy highway drivers from the perspective of human drivers.
My 2022 car when operating in "Adaptive Cruise Control" mode (as a simple robot) on the highway responds as it should for safety to a slowing car ahead, by sacrificing speed for distance to maintain adequate calculated braking distance. This "correct" action is the opposite of what humans do, who generally sacrifice distance to reduce speed as little as possible. As such my good robot driver slowing down more than a human creates an unexpected and potential dangerous situation behind me and at a minimum annoys the driver following me.
What happens next as traffic speeds up is my robot does not speed up as quickly to avoid aggressive acceleration and so to the driver behind me I am annoyingly not keeping pace.
Note that in both cases, the robot is driving correctly and safely. Should the programmers teach it to drive like a human or someone's grandmother and stay in the right hand lane? Would you like to be in that car?
Worse, when the highway curves to the left while an exit is straight ahead, my car sometimes misinterprets the car in front of me slowing down to exit as an abruptly slowing car on the highway. In response, my car breaks heavily and unexpectedly, startling the driver behind me who is hopefully paying attention and who sees no reason or warning for me to brake. Rage? I have nearly been re-ended twice in this scenario.
The more robots, the more grandmothers there are on the highway which I submit is not a better safety environment. Perhaps a bumper sticker would be appropriate. "Don't Blame Me, My Robot is Driving"
I think the bigger thing is that human drivers naturally tend to drift to the other side of the lane of a car when passing to get a bit more distance. Robots don't do that. Which is probably not the right course of action when other human drivers are still on the road.
Yeah, no
Self driving vehicles are inherently privacy invasive. That ought to give pause of itself.
Also, when a driver causes a homicide we understand that it's a failing of that individual. However, when an autonomous vehicle causes a fatality it's because of what a programmer or engineer thought was an appropriate trade off when training the engine.
They are different categories of both culpability and morality.
Not absolutely against, but it's not just a straight utilitarian issue.
It's worse than that though because any given failure may not be traceable to a specific decision made by a specific set of people. The systems are black boxes by design - a giant pile of training data is ingested, resulting in a bunch of inscrutable neural net weights. It's usually going to be impossible to trace some action of one of these systems back to some particular bit of training data. Why did ChatGPT say that totally made up thing? No one knows, and it certainly wasn't in the training data.
Surely Alphabet, Inc. will be transparent with the data - and inherent liability it corroberates and will not use it's golliath power to lobby for laws that prevents normal people harmed by this technology (an acceptable sacrifice, on the net! will save thousands of lives, no argument!) from seeking recourse that would otherwise bankrupt it durring the adoption phase
He also said that there is a human sitting somewhere, explicitly not in the car, who can over ride the car.
So humans are involved just not there to live with the consequences of their actions. Physically anyway.
The new drone pilots.
A lot of that can probably just be handled by insurance.
How sad to see such a technologically illiterate article in a magnificent journal like The Free Press. The Free Press has an admirable moral compass, but it is weak in the realm of real world technology. How absurd is the proposition that “X percent (pick a number) of traffic accidents are caused by human drivers. Therefore eliminating human drivers will reduce traffic accidents”? The pursuit of self-driving technology has yielded some great driver-assist systems, but the claim of self-driving, except under the most constrained conditions, will yield only carnage. Free Press, you can do better than this.
In the late 'twenties of the last century, a barnstormer crossed the Midwest, landing at farms along the way and giving rides to locals. An old farmer, driving his horse and flatbed wagon, brought his young son to see the spectacle.
He approached the pilot rather skeptically and eyed the machine up and down. Finally he rubbed his chin and said, "What good is it?"
The barnstormer pointed to the young son standing by the old man's side and said, "What good is HE?"
"Why, he's just a child now, but someday he will be a man!"
"Exactly."
That's like looking at cell phone's in the 1980's and assuming they were always going to be too big and expensive to be practical.
Self driving cars aren't quite there yet, but when fully rolled out are sure to be orders of magnitude safer than human drivers.
Author: "Anyone rooting against self-driving cars is cheering for tens of thousands of deaths, year after year."
Really... you are "cheering for death" if you don't agree with the move toward less individual freedom and more government/corporate control of human life? This kind of hyper-emotional attempt to shame those who hold another opinion tells me that I can disregard the rest of this guy's argument.
I'm not a luddite. But the safety justification better not open the door to taking away my autonomy. I like driving. Take a lesson from unusable gas containers, low pressure toilets and showers.,
Let's also take a lesson from the humble traffic circle. Reduces accidents by 66%, fatalites by like 92%, emissions by like 40% because your car is least efficient from 0-10mph and you can kinda keep rolling through it. And pedestrian fatalites. And stress. And you can use 1 lane instead of 2-3 at the same efficiency.
If this was purely data-driven, THINK OF ALL THE LIVES THIS WILL SAVE we'd tear down every stop sign and traffic light in the country and have nothing but traffic circles because we can do that in 6 months instead of waiting .... how many years now? For waymo to get thier shit on the road and save us from ourselves.
I detest traffic circles
You know it's going to happen. You're a bad driver, don't you see? You shouldn't be allowed on the road like an animal. Or you're literally murdering people.
"The innocent self-driving car was burnt to a crisp."
Tip #1 for reaching people on the other side of this issue: do not anthropomorphize or ascribe moral value to technology. It smells weird and gets the hackles up. Reserve moral categories for the humans creating and using the technology.
A normal car is a physical extension of it's owner that Americans feel a deep emotional connection to, so I could see how the author could fall into this mistake.
A Waymo car is the physical extension of Alphabet, Inc. - and Google has in 70 billion ways not covered itself in glory this month on the "trust us" front.
Honestly I wish more tech companies would offer us physical manifestations that we can publicly immolate.
“ Anyone rooting against self-driving cars is cheering for tens of thousands of deaths, year after year.”
I guess I'm cheering for tens of thousands of deaths, year after year, then.
Utilitarianism is evil. Automating everything and delegating to AI takes away human agency. You'll have to rip my steering wheel from my cold dead hands.
"And contrary to the view that driverless cars are being deployed unilaterally by tech billionaires, the people’s representatives—government officials—gave Alphabet-owned Waymo a license to operate."
Tell me you don't see the irony in this sentence.
I love FP but they need a better shill filter
I appreciate reading stuff from all sides, at least the FP is being open and not shutting down things its readers may not agree with.
Typical leftist Utopian soft totalitarian thought process.
People just shouldn’t be allowed to do things the Oligarchy deems dangerous. It’s for people’s own good. It’s for the innocent children. It saves money. Exceptions for Nomenklatura of course.