If you've ever taken an armchair interest in philosophy or ethics (or, let's be honest: if you've ever taken a bong rip in a college dorm room,) you've probably heard of the "trolley problem."
Usually attributed to philosopher Phillipa Foot in 1967, it goes like this: five people have been tied to the tracks by some Snidely Whiplash-style cartoon villain, and an unstoppable runaway trolley is about to run them over — unless you pull a lever that will divert the trolley onto another track.
The only problem? Snidney is a prepared serial killer who has covered his bases, and he's tied another innocent bystander to that track, albeit just the one. So at least one person is going to lose their life today — and it's up to armchair ethicists everywhere to determine who should die.
The trolley problem, of course, isn't really about trolleys; it's about the ethical question at the heart of (frankly too many of) our policy choices, particularly those that govern our transportation policy: how do we decide whose death is acceptable? And more important: why are we willing to pull that proverbial lever to save someone's life, even if it means simultaneously ending some else's, instead of just creating a future where no one dies at all?
In a recent interview at TechCrunch's Disrupt Summit, Waymo co-CEO Tekedra Mawakana made it all too clear that she is more than prepared to pull that theoretical lever, when she was asked, "Will society accept a death potentially caused by a robot?"
Mawakana replied: "I think that society will ... We don’t say ‘whether’ [our vehicles will be involved in fatal crashes]; We say ‘when.' And we plan for them.”
The "plan," the CEO implied, involves pulling vehicles out of service temporarily for retests. And to be fair, Waymo "seems heads and tails above its competition because its rollout into city streets has been extremely slow and deliberate," as Sharon Ardarlo wrote for Futurism.
Unlike robotaxi competitor Cruise, a Waymo has never dragged a pedestrian down the street; unlike Tesla, it doesn't market so-called "Full Self Driving" technology to everyday consumers, despite the fact that it's ... not (which might explain why Tesla drivers were ranked most likely to be involved in a crash of any car brand).
Still, Mawakana's response echoes a utilitarian attitude that has been shared by countless AV proponents before her, along with a disappointing proportion of the traveling public at large. These proponents argue that if AVs will someday avoid more crashes and save more lives than human drivers — and, indeed, in certain limited, near-perfect conditions, they already can — it's worth pinning our hopes for our safety future (and more than $100 billion and counting) to the promise of them.
Pull the lever. Save the five people tied to the tracks. Let the single person die, and pat ourselves on the back for minimizing our losses.
Of course, Mawakana isn't the only person in the AV industry to muse about the limits of the poisoned moral and political environment into which her vehicles are deployed. Trolley problem-type conundrums are among the most frequently lobbed at AV CEOs like her; they're even more relevant to autonomous vehicle programmers, who are directly tasked with instructing their on-board AI how to behave in ethically fraught roadway situations.
In a 2023 interview for Futurity, Chris Gerdes, the co-director of the Center for Automotive Research at Stanford, took on an AV-specific edition of the trolley problem when he explored a theoretical situation in which a non-automated vehicle like a bicyclist was "blowing through a red light or turning in front of the AV."
"Blowing a red light" on a bike, for the record, is perfectly legal in nearly a dozen states if a cyclist slows down and determines it's safe to proceed. Known colloquially as the "Idaho stop," the "stop-as-yield" maneuver has actually been found to reduce in crashes in studies, because it reduces unnecessary interactions between drivers and riders and helps cyclists get ahead of car traffic where they are more visible, rather than maximizing their chances of landing in a motorist's blind spot.
In Gerdes' example, though, he argued that a cyclist practicing the Idaho Stop in a state where it's not legal would likely "violate [the cyclist's] duty of care to the AV" by violating that law. Moreover, he claimed that violation would justify the AV's decision to plow into the cyclist and kill them, at least if it meant not swerving into the opposite lane and striking, say, a car with five people inside.
"Why would it be OK [for an AV] to avoid a bicycle by swerving an automated vehicle out of its lane and into another car that was obeying the law?" Gerdes wondered. "Why make a decision that harms someone who is not part of the dilemma at hand? Should we presume that the harm might be less than the harm to the bicyclist? I think it’s hard to justify that not only morally, but in practice."
What Gerdes and Mawakana's analyses both miss, though, is that there's a simple, low-tech way to avoid the AV striking either the oncoming car or the cyclist. And that's removing the AV (and all other cars) from the equation entirely — by putting the cyclist in a protected lane that no car can physically enter.
Whether their preferred traffic violence "reduction" strategy is heavy on autonomous vehicles or constant, violent policing or something else entirely, too many Americans consciously or unconsciously treat road safety as a zero sum game. And that's because we've been conditioned by automobility (and the corporations that profit from it) to believe that "fewer" lives lost is an automatic net positive — and if the net total deaths at the end of the day is more than zero, well, at least that's better than the roughly 40,000 human beings who are slaughtered every year on our roads.
This utilitarian attitude is comforting, because it provides clear, black-and-white answers to the murky "trolley problems" we face on the U.S. transportation system every day — even as trolleys themselves increasingly vanish.
If the AV swerves into the bicyclist to avoid the van full of Girl Scouts on the way to sell cookies, utilitarianism says that's a clear net good, rather than tragedy that could have been avoided for the Scouts and the cyclist.
If the traffic modeler says widening the highway will decrease crashes by a certain impressive percent — even if those models are entirely untested, and ultimately prove to be wrong — then a myopic, crash-focused utilitarianism will demand that we widen it, even if it means pouring pollution onto surrounding communities and killing people slowly with heart disease and cancer, along with the countless other, quieter ways that highways kill communities and eventually the people in them.
The trouble with aspiring to "fewer" deaths — rather than demanding zero — is that utilitarianism conditions us treat our roads like a trolley problem, in which we have only a narrow range of safety choices, all of which result in at least some level of slaughter. But in reality, our roads are a complex system, and if we make enough layers of that system safe, we can end road deaths entirely.
Much like we can just build the protected bike lane that renders Gerdes' cruel thought experiment moot, we don't have to pull the lever on the proverbial trolley switch at all; we can rush to untie all six people from the trolley tracks and simply set them free.
If there isn't enough time to do that, we can require emergency braking systems on trolleys that make crashes impossible at least at slow speeds, much like we should on all cars and trucks.
We might even try to get to the root causes of why the hell there are all of these runaway trolleys all over town, much like we badly need to scrutinize our threadbare national patchwork of AV regulations.
At the very least, we can stop the Snidely Whiplashes of the world from buying bulk quantities of rope, like some states are proposing to stop super-speeders by putting speed limiters on their vehicles. And we can string all these solutions and more together to make it impossible for even the most determined of supervillains to commit a horrific trolley-assisted mass murder.
Most important, though, we can recognize that our transportation future isn't a runaway train we're powerless to stop. Mass automobility — yes, including the the AV-dominated version — is a choice we have had collectively imposed on us by the powerful. We can choose another system that prioritizes safety proactively, alongside all the other benefits of a car-optional world.
