Even After $100 Billion, Self-Driving Cars Are Going Nowhere

Even After $100 Billion, Self-Driving Cars Are Going Nowhere

Max Chafkin

Our driverless future is starting to look so distant that even some of its most fervent believers have turned apostate. Chief among them is Anthony Levandowski, the engineer who more or less created the model for self-driving research and was, for more than a decade, the field’s biggest star. Now he’s running a startup that’s developing autonomous trucks for industrial sites, and he says that for the foreseeable future, that’s about as much complexity as any driverless vehicle will be able to handle. “You’d be hard-pressed to find another industry that’s invested so many dollars in R&D and that has delivered so little,” Levandowski says in an interview. “Forget about profits—what’s the combined revenue of all the robo-taxi, robo-truck, robo-whatever companies? Is it a million dollars? Maybe. I think it’s more like zero.”

“Humans are really, really good drivers—absurdly good,” Hotz says. Traffic deaths are rare, amounting to one person for every 100 million miles or so driven in the US, according to the National Highway Traffic Safety Administration. Even that number makes people seem less capable than they actually are. Fatal accidents are largely caused by reckless behavior—speeding, drunks, texters, and people who fall asleep at the wheel. As a group, school bus drivers are involved in one fatal crash roughly every 500 million miles. Although most of the accidents reported by self-driving cars have been minor, the data suggest that autonomous cars have been involved in accidents more frequently than human-driven ones, with rear-end collisions being especially common. “The problem is that there isn’t any test to know if a driverless car is safe to operate,” says Ramsey, the Gartner analyst. “It’s mostly just anecdotal.”

It’s mainly juridical issues that limits the usage of self-driving cars. It can not be proven they are safe and there’s nobody to put responsibility to when it fails.

It’s mainly juridical issues that limits the usage of self-driving cars. It can not be proven they are safe and there’s nobody to put responsibility to when it fails.

You really think this is merely juridicial? ->

Tesla car in Beta full self-driving mode almost rams cyclist (+ video)

Top Tesla engineer confesses he staged an inaccurate self-driving video at Elon Musk’s bidding

Tesla Officially Admits Its Self Driving Cars Cause Crashes With A Massive Recall

It's pretty obvious the technology was oversold, along with a lot of machine "learning" and now investments are dropping.

Companies oversell the self-driving capabilities of their cars, with horrific outcomes

Even After $100 Billion, Self-Driving Cars Are Going Nowhere

Max Chafkin

Google has already got some working driverless taxi's on the road since 2019, more are here now, and there are more coming. The same has just begun to happen in China.

Yes, there are lot's and lots of issues to sort out including separating different types of traffic, new traffic laws, installation of new electronic signage and markings, newer accurate GPS, 5G+ for remote drivers/assistance, and centralised memory/learning... it is an enormous change. I'd say that the UK could have significant numbers of driverless taxi's on the road during the 2030's - eliminating individual vehicle drivers is a real competitive advantage for taxi operators. In the future I just expect more and more working class people will choose not to own outright a car which sits at work, and at home for the overwhelming majority of it's life.

You really think this is merely juridicial? ->

Tesla car in Beta full self-driving mode almost rams cyclist (+ video)

Top Tesla engineer confesses he staged an inaccurate self-driving video at Elon Musk’s bidding

Tesla Officially Admits Its Self Driving Cars Cause Crashes With A Massive Recall

It's pretty obvious the technology was oversold, along with a lot of machine "learning" and now investments are dropping.

Companies oversell the self-driving capabilities of their cars, with horrific outcomes

yes I'm familar with 'sensationalism' in the medias. But human driving is not flawless either. In fact AI assisted driving is mainstream in most new midrange and above cars nowadays. They help you to keep you car within the driving lanes and keep the distance to other cars on the road. It's making driving a lot safer.
Investments in AI are predicted to climb globally to $232 billion in 2025 according to a recent report from KPMG. It will replace millions of workers in the next decade. Here's a list of all the exams chatGPT 4 has passed so far. List: Here Are the Exams ChatGPT Has Passed so Far (businessinsider.com)

yes I'm familar with 'sensationalism' in the medias. But human driving is not flawless either. In fact AI assisted driving is mainstream in most new midrange and above cars nowadays. They help you to keep you car within the driving lanes and keep the distance to other cars on the road. It's making driving a lot safer.
Investments in AI are predicted to climb globally to $232 billion in 2025 according to a recent report from KPMG. It will replace millions of workers in the next decade. Here's a list of all the exams chatGPT 4 has passed so far. List: Here Are the Exams ChatGPT Has Passed so Far (businessinsider.com)

That article is just more sensationalism? Also passing an exam isn't like driving. Investment is just like the $100B for driverless cars that achieved so little, it isn't a sign of definite progress.

I agree programs that assist humans will be more common but the public is going to get less tolerant of the hype as errors mount in all sorts of machine "learning" applications - errors that are difficult to predict in advance by the very nature of these kinds of programs. This issue will only be compounded when people try to deliberately mess with these programs.

This Artificial Intelligence (AI) Research Proposes A New Poisoning Attack That Could Trick AI-Based Coding Assistants Into Suggesting Dangerous Code

ChatGPT’s Most Charming Trick Is Also Its Biggest Flaw

Google has already got some working driverless taxi's on the road since 2019, more are here now, and there are more coming. The same has just begun to happen in China.

Yes, there are lot's and lots of issues to sort out including separating different types of traffic, new traffic laws, installation of new electronic signage and markings, newer accurate GPS, 5G+ for remote drivers/assistance, and centralised memory/learning... it is an enormous change. I'd say that the UK could have significant numbers of driverless taxi's on the road during the 2030's - eliminating individual vehicle drivers is a real competitive advantage for taxi operators. In the future I just expect more and more working class people will choose not to own outright a car which sits at work, and at home for the overwhelming majority of it's life.

I think it is possible to have a driverless car, but the way the current programs just try to bound uncertainty by putting in more and more data...that I don't think is workable for driving.

There are ideas like using specific programming paradigms to help build what I would call actual AI rather than what machine learning does. Like this article on functional programming mentions (applogies in advance for the green background color he uses):

Haskell is in many respects a very safe language. This is why it’s trusted by large financial institutions, large data science firms, and even by a company working in autonomous flight control. When your code cannot have arbitrary side effects, it is far easier to prevent it from crashing. It is also easier to secure a system (like an IoT device) when you can prevent leaks from arbitrary effects. Often these techniques are present in Haskell but not other languages.

The field of dependent types is yet another area where we’ll be able to add more security to our programming. They’ll enable even more compile-time guarantees of behavior. This can add a great deal of safety when used well. Haskell doesn’t have full support for dependent types yet, but it is in the works. In the meantime there are languages like Idris with first class support.

Of course, when it comes to AI and deep learning, getting these guarantees will be difficult. It’s one thing to build a type that ensures you have a vector of a particular length. It’s quite another to build a type ensuring that when your car sees a dozen people standing ahead of it, it must brake. But these are the sorts of challenges programmers will need to face in the AI Native Future. And if we want to ensure Haskell’s place in that future, we’ll have to show these results are possible.

I do think trying to use something like Haskell or Idris will lead to issues of reaction time without hardware that allows for "bare metal" functional programming.

All that seems very much in its infancy, and not sure any of the current questionable progress the article in my opening post goes into will be transferable.

In fact AI assisted driving is mainstream in most new midrange and above cars nowadays.

Yes, but AI assist is an utterly different type of technology and the driver is always there to override it.

Think of automatic parking. The car is moving at a very slow speed, and can safely stop if anything gets in its way.

More sophisticated AI-assist is dangerous because people lose concentration while being driven, and the effort to regain concentration just when things are going pear-shaped is massive.

Driverless cars would be very useful if they could cope with everything we can - and maybe more:

Poor visibility.

Roadworks.

Encountering an accident.

Assorted animals in the road.

Children behaving carelessly in various ways.

Encountering drunk people, or people with various disabilities.

Various malfunctions in the car.

Very narrow roads, such as you can find in the UK.

Roads with pot holes.

Police trying to direct traffic for various reasons.

Etc.

David

Cruise Cars Crash Into San Francisco Muni Bus And Tangle In Fallen Trolley Wires

Brad Templeton

A Cruise Robotaxi, with nobody aboard, has rear ended a San Francisco Muni bus on March 23. Nobody was injured and damage to the bus was modest, but more severe to the front bumper of the Cruise. There were no passengers on board the vehicle — Cruise does not carry public passengers during the day, though GM employees are permitted to ride.


Cruise issued a very minimal statement saying:

"Yesterday one of our vehicles made contact with the rear of a Muni bus. No injuries were reported, and there were no passengers present in the Cruise AV. We are investigating the incident and will take follow up actions if appropriate."

Cruise confirms the vehicle had no safety driver and was in autonomous mode.

It gets worse for Cruise. During the major high-wind storm earlier in the week, a Cruise vehicle drove through “caution” tape closing off a road where overhead trolley wires had fallen. The car then hit the wires, for which power had been shut off.

Both incidents look very bad for Cruise, and they could easily have been much worse if there had been passengers in the vehicles, or if the power lines had been live.