Tech

Elon Musk Claims New Government Traffic Safety Advisor Is ‘Biased Against Tesla’

A Duke University engineering professor critical of self-driving car companies has been named a safety advisor for an auto regulator.
Elon Musk
Picture alliance via Getty
Screen Shot 2021-02-24 at 3
Moveable explores the future of transportation, infrastructure, energy, and cities.

The National Highway Traffic Safety Administration (NHTSA) has hired Missy Cummings, a Duke University engineering and computer science professor, as a senior advisor for safety, Reuters reported. It is an interesting hire for an agency tasked with overseeing and potentially regulating the autonomous vehicle industry, as Cummings has been one of the more skeptical experts on the potential for self-driving cars to actually exist. It also lends further credence to NHTSA-watchers who believe the agency is positioning itself to take a more active role policing the sector including semi-autonomous driving software widely available on cars today.

Advertisement

But that is not why we are here today. You are reading about this obscure government agency hire because it has made Tesla fans very mad online, for reasons that are both simple and complex. 

The simple reason is Elon Musk said something about it, so his fans have responded accordingly. “Objectively, her track record is extremely biased against Tesla,” Musk tweeted.

The complex reason is the answer to the question: What is that track record?

Cummings is a former Navy fighter pilot who has researched “human factors” engineering for decades. In a 2010 MIT news profile—where Cummings used to work—she explained she was partially motivated to enter this field after watching a friend die landing a fighter plane on an aircraft carrier. She later went on to study, among other things, how to structure air traffic control systems so the people monitoring them don’t get bored, complacent, or trust a flawed automated system too much, any of which could result in catastrophe. In the years since, human factors research has become deeply relevant to semi-autonomous and fully autonomous cars. 

As you might imagine from someone with equal parts respect and caution for what automated technology can do, Cummings has frequently been critical of the hype cycle surrounding autonomous vehicles and artificial intelligence in general. Her criticisms are not limited to any particular company or technology, but extend to the field as a whole. 

Advertisement

She has argued the entire concept of using machine learning to create an artificial intelligence system in safety-critical settings like self-driving cars is flawed. Humans are generally good drivers because we’re capable of using both bottom-up and top-down reasoning to interpret what we see and hear. For example, if snow is partially covering lines on the road, we can infer the lines extend underneath the snow. Computers cannot do anything like that; they process by brute force billions of data points and cannot infer anything. Cummings argues this is far too brittle a process to be relied upon when people’s lives are at stake.

A few months ago, I reached out to Cummings about a story I was working on regarding self-driving cars in cities. I asked her, “if you were cycling along a road, would you rather have a Waymo driverless vehicle next to you or a randomly picked licensed human driver?” I asked about Waymo specifically because it is widely regarded in the industry as the safest, most advanced, and responsible self-driving car company. She said she’d rather have the human. 

Advertisement

“I actually bike quite a bit, and even though humans are horrible drivers, I know what to expect from this set of bad drivers,” Cummings replied. “At the present time, I have no idea what AVs might do and this is where the real problem lies.”

These are fair criticisms worthy of intense public debate as technologists seek to reshape the world in which we live and how we move about that world. Instead, some Tesla fans have ignored all of that and focused on a number of other issues, including the fact that she sometimes sends some fire tweets, an especially odd criticism from a culture that idolizes Elon Musk, who has been sued because of his tweets multiple times

But there are two specific issues of some substance being raised. The first is that she co-authored a paper regarding driver-assist systems, the most prominent of which is currently Tesla’s Autopilot. Even as Tesla fans recount the episode, it is benign at most. The paper included details about a Tesla crash in a Houston suburb where the police initially said with certainly no one was behind the wheel at the time of the crash. The paper was updated to remove that detail since the National Transportation Safety Board investigators cast doubt on that narrative, which has yet to be fully resolved as the investigation remains ongoing. This instance has been trotted out as evidence of her “anti-Tesla” bias.

Advertisement

Second, and more frequently cited, is Cummings’s role on the board of a company called Veoneer, for which she receives stock compensation. This company makes lots of things, one of which is LiDAR, or Light Detection and Ranging. It’s like radar but with lasers to form more detailed three-dimensional images of the world. And it is a trigger word for Tesla fights.

Tesla is virtually alone among companies with self-driving aspirations in not using LiDAR. Musk and Tesla fans say they don’t use it because it’s not necessary and self-driving can be achieved with camera-based systems alone. Others say this isn’t true, that cameras have inherent limitations, and that detailed LiDAR images are needed for computers to gather enough data to properly interpret the world around them. Tesla skeptics also sometimes argue the company doesn’t want to use LiDAR because it would mean all the cars they have sold to date without it cannot achieve self-driving with a simple software update, something Musk has promised at investor presentations would be possible for years, including the claims that Tealas could one day act as fully autonomous robotaxis and earn Tesla owners money. If all of that ends up being demonstrably untrue because the cars aren’t equipped with the necessary hardware, Tesla could possibly be subject to lawsuits.

Advertisement

As a result of this fight, LiDAR companies are perceived as villains in the Tesla Cinematic Universe, a series of interactive installments of Us Versus Them narratives thousands of people around the world have tied to their online identities. Back in the day, Big Oil and Ford were Tesla’s biggest enemies because an electric future threatened their existences. Now that Ford says it’s all in on an electric future, LiDAR companies are the new heels. And, by serving on a board of a company that makes LiDAR, Cummings fits neatly into that narrative.

The only flaw with this narrative is Cummings’ actual stance on LiDAR. “Lidar is not the answer,” she tweeted in August. “NO company will ever get to anything like full self driving without a complete rethink of reasoning under uncertainty.” 

Which brings us to the main reason Tesla fans have a problem with Cummings. It’s not because Cummings is critical of Musk or Telsa about any one thing, but because she is not on board with what Musk is selling. Musk is a futurist, and Cummings is more of a realist. For example, in a 2019 paper on the use of automated technologies in military settings, she wrote, “I find these debates filled with a lack of technical literacy and emotional rhetoric, often made worse by media and activist organizations that use fear to drive exposure and funding.” She says automated technologies perform best in “a very narrow task with little to no uncertainty in their environments,” such as maintaining a lane on a highway as long as there is good lighting and visibility. This is a fundamentally competing worldview to Musk’s.

One of the tweets Tesla fans are citing the most in their opposition to Cummings is one she sent in 2018, responding to a Deadline article about a new documentary on the threat of “artificial intelligence.” The documentary itself, as explained by director Chris Paine, explored the question “Can we control what we’ve created,” a line of inquiry AI skeptics have criticized as implicitly buying into the hype machine and thereby lending AI more credence than it warrants. 

For her part, Cummings called it an “exceedingly alarmist view of overly privileged, narcissistic and insecure white men who need to assuage themselves that they are part of some greater good and here to save us from the folly that they created. The only killer robot out there is @elonmusk’s Tesla.”

Cummings is such a flashpoint to Tesla fans not because she owns stock in some company you’ve never heard of. (For the record, I believe she should have to resign from the board and divest of that stock before assuming a role in a vehicle safety regulatory agency. Cummings declined to comment for this article.) They dislike her for the same reason they pile on anyone else who criticizes Tesla and Musk. Musk is telling his fans and followers technology can make the world better by having computers assume dangerous tasks previously done by humans, and that this will usher in a transhumanist utopia. Anyone who argues otherwise isn’t just wrong to them, but a threat to the better world Musk is trying to build.