Opinion: Elon Musk’s autonomous vehicle forecasts are not just too optimistic — they’re wrong


Autonomy is supposed to make trips to Tesla’s TSLA,
-6.42%
affordable and safe planned robotaxis. Musk predicted that his company would “far exceed the level of human safety … ultimately a factor 10 safer than a human measured by the likelihood of injury.”

Don’t bet on it.

If another Musk prediction turns out to be too optimistic, no one will be surprised, let alone Musk. But the problem here is not over-optimism about the pace of improvement. The so-called “autonomous vehicles” not only take longer to develop than expected. They are a dead end.

AV skeptics have a decade of failed promises. The response from AV advocates is that the forecasts were too optimistic about the pace towards the destination. The pace of progress may be disappointing, they admit, but we will get there.

But we have much more than a decade of experience to learn from, and that experience reveals persistent biases that inflate expectations.

We have been promised that engineering will almost eliminate motor vehicle accidents for 90 years. In 1934, the leading American expert on the subject, Miller McClintock, promised that state-of-the-art highways could be “built in such a way that accidents would be impossible.” With access controls, grade separations, middle stripes and shoulders, he wrote, “infallible highways” are possible. McClintock estimated that these roads would reduce fatalities by 98 percent, enough to justify a promise of “complete permanent safety.” In the press, McClintock’s claims were widely accepted as facts. In 1937, the Chicago Tribune made the “death-proof highways” the number 1 board of its publishing platform.

Restricted access roads are much safer than other roads, but not as safe as McClintock promised. If he had been right, the death toll today would be too small to give AV promoters their favorite moral argument.

Musk promises to make spectacular profits using an incalculably more sophisticated technology – a technology that takes full control of the human driver. This obvious fact, however, hides the same elementary error in both predictions, and in many others over the intervening decades. To reach their conclusions, the manufacturers of promises subtract all the dangers that make the status quo dangerous, without adding all the important new dangers that innovation introduces.

On restricted access roads, most of the new hazards are due to the speeds that these roads invite. If no driver exceeded 20 mph, the deaths in them would fall close to zero. But highways justify their cost by allowing them to drive much faster, at speeds that can be deadly. Similarly, AV technology can make them much safer, but the faster the vehicle travels, the less technology it can make.

Sensors can detect almost instantly, but like humans, AVs may need time to interpret unusual objects with high reliability, and they need to track moving objects over time to predict how they will behave. At low speeds, these problems are manageable. A slow AV has time to interpret, follow, predict, and respond, and if it still fails, a low-speed collision may be less. But to attract a large number of paying motorists, AVs would have to offer speed. And at high speed, AV errors will be much more frequent and much more lethal.

The comparison with the “infallible highways” has more to offer. People driving on restricted highways must drive on normal roads and streets to get there and back. The benefit of road safety cannot be extended everywhere, and safer roads can increase casualties on the surrounding roads just to attract more driving.

Island Press

Similarly, there is no direct path to a future unmixed AV in which all vehicles on the roads are automated. There must be many years of mixed AV and conventional vehicles. Conventional cars can be connected to a network, making it easier to monitor AVs, but mixed traffic will continue to be so difficult that AVs will have to be dangerous or slow.

If AVs are slow, this will lengthen the transition (why go with a slow AV?) And introduce new safety hazards as frustrated human drivers try to pass slow AVs.

Eventually, the AVs will continue to disappoint because they are misnamed in a way that inflates what we expect of them. “Autonomous” predisposes people to think of AVs as impartial hyperrational beings with an “I” who can drive and enough discretion to exercise autonomy. The misperception is evident in Musk’s choice of words: “autonomous driving … this is better than human.” When developing an AV, the goal of engineers is to make sure that everything the vehicle does is what its developers and human operators want it to do. The unpredictable behavior of the vehicle must be removed. The goal in the development of AV is to ensure that humans determine as thoroughly as the vehicle works that the vehicle itself can have zero range.

AVs are in fact human-driven vehicles, even though the operators are not in the vehicle, and they control it through programs instead of a direct physical interface. These human operators must be biased; otherwise, they will have no paying human customers. Except for niche applications or loss-making services (such as Alphabet’s GOOG,
-1.29%

GOOGL,
-1.34%
Waymo, and perhaps Tesla’s possible robotaxis), shouldn’t value safety so much that the vehicle never exceeds 20 mph. They should also value the customer experience rather than the preferences of others outside the vehicle; if they don’t, they will lose their customers.

These human biases are business imperatives and have a security cost that prevents the kind of security gains that Musk and other AV promoters have predicted. AVs can’t become commonplace until they can get more revenue than they cost. And they can’t do that and prioritize security over customer satisfaction.

AV promoters have not been too optimistic about the pace of AV development. They have been wrong about what AVs offer.

Peter Norton is an Associate Professor of History in the Department of Engineering and Society at the University of Virginia in Charlottesville and author of “Autonorama: The Illusory Promise of High-Tech Driving.”

Now read: Elon Musk said ESG was a scam: Did Tesla’s boss do investors a favor?



Source link

Leave a Reply