The Greenwashing of Robotaxis

The benefits promised for robotaxis might not work out as promised, and might be easier to get in other ways.

According to Autonomous Vehicle industry boosters, AVs such as robotaxis are going to save the world! We’ll have improved safety, reduced emissions, better transportation equity, better accessibility, less congestion, and more efficient energy use. However, typical claimed benefits range from spouting hype to repeating debunked disinformation (for example, the thoroughly debunked claim that 90%+ of crashes are caused by human error).

We need to ask the hard questions. Are AVs really the only way, or even the best way to achieve these goals? And even if AVs deliver as promised, will we be better off? Or are the claimed environmental, equity, and other benefits just a greenwashing exercise for robotaxis?

Improved safety

Data to date is insufficient to know how fatalities and severe injuries will turn out for AVs. That’s just not me saying this — that is Waymo’s own research team saying this. We are years away from knowing how fatality rates will turn out.

But we already know that other much less exotic vehicle features can dramatically improve safety. Automated Emergency Braking is amazing stuff! And other countries have done much better at improving safety than the US in the last decades. We don’t need AVs to improve road safety. There are proven ways to get us there, with some of them already happening. If we want more safety we just need to get serious about using known approaches and fostering the evolution of existing capabilities. Waiting for AVs that might eventually operate at enough scale to move the needle is the lazy way out — which might not even work. If AVs are no worse than human drivers they might be viable despite not delivering on the safety hype.

Robotaxi deployments can, however, dramatically decrease safety even if AVs are proven to be just as safe as a good human driver. That happens when they draw passengers away from mass transit, which can easily be safer than human-driven vehicles. (And which can likely be made even safer fairly easily using existing technology if that is a priority for us.) On a per-trip basis, drawing passengers away from mass transit could easily reduce net transportation system safety and disincentive other safety and transportation investments.

Improved equity & accessibility

AV supporters talk about solving food deserts and improving transportation accessibility to the disabled. But ride hail with human drivers was supposed to do that — and had limited success. A lot of this is because businesses are incentivized to chase money rather than provide equitable service (because, capitalism). Regulatory capture prevented effective enforcement of equity and accessibility promises made by ride hail drivers. Why in the world would we think that AVs will turn out any different?

Beyond that, without a human driver there will be issues for those who need help with vehicle boarding, getting groceries to a front door for those not able to climb their front steps, and so on. That’s assuming the promised wheelchair-accessible vehicles actually get deployed, which is still very much a work in progress.

Sure, it could happen, but I’ll believe it when I see California PUC (and other states) require companies to deliver on equity promises used to garner their operating permits. Not holding my breath, but would be delighted to be proven wrong.

Yes, eventually AVs might be cheaper than ride hail, and thus be more affordable. But cheaper rides are quite likely to further degrade the viability of public transit. This could well leave the least affluent, who cannot afford even cheap(er) robotaxi rides, even worse off than before.

Less congestion

So let me get this straight: the AV industry is going to make robotaxi rides cheaper, which will increase demand, but somehow reduce congestion. Did I get that right? Seriously? How can people say stuff like this with a straight face??

The bit about cars being underutilized assets that are dormant in a parking garage all day is a non-sequitur here. Congestion is about the number of cars on the street. Ride hail business models tend to increase that with deadhead miles. Individual cars require parking, but they don’t have deadhead miles on roads. The robotaxi fleet used during morning rush has to be somewhere all day. If it’s parked then it is no different than commuter cars. If it is on the streets that is congestion. If it heads out of town to the suburbs that doubles miles per day (two round trips per vehicle). Not seeing a clean win here.

Did someone say “shared vehicles?” Shared vehicles aren’t going to work unless someone cracks the nut of people just hating to share vehicles. Shared vehicles are a failed dream for ride hail. It won’t be any different for robotaxis. Especially since one of the biggest advantages people love about robotaxis is there is no driver to share the vehicle with (for example, for personal safety, which is more of a mixed proposition than many have considered). You don’t want to share with a driver but you are happy to share with a passenger who is even less accountable than a driver in the cabin? I don’t think so.

If you want to get rid of parking to improve the city that might get interesting. Have a plan for what the robotaxis do between rides and how commute peak is handled. But that is a generational change that requires a massive re-think of everything. Cool idea. Want to hear more about how it will work and how the transition will happen. Let’s see how it plays out in European cities going down that path.

But simply dumping robotaxis onto city streets with no other changes seems likely to make congestion worse for the foreseeable future, not better. If you want to introduce metrics such as a PUC requiring that a substantial fraction of AV trips must connect to public transit, well then maybe it gets interesting. But I don’t hear that conversation going on.

Reduced emissions & energy use

This is an EV talking point. The AV narrative is that robotaxis will increase fraction of EV rides. But so will equipping ride hail drivers with EVs in our current system.

Worse, the computers on AVs use a significant amount of electricity. As do the data centers used to train them. So if you care about carbon emissions, a ride hail EV might well outperform an AV-driven EV. (There is a lot of analysis detail needed here — I’m just pointing out that this is not a slam-dunk AV win. Especially since machine learning retraining will be a life-long requirement to deal with evolving environments.)

Moreover, what if more people are using cheaper AVs instead of public transit? An EV robotaxi might have desirable energy use compared to a non-EV robotaxi, but public transit can easily win in this area per passenger-trip.

Summing up

You can have improved safety, reduced emissions, better transportation equity, better accessibility, less congestion, and more efficient energy use without autonomous vehicle robotaxis. It is questionable whether robotaxis will break even on any of these metrics at the transportation system level, let alone compete with non-AV approaches that are better than the status quo.

Autonomous vehicles might provide significant benefits (for example, freight movement; improved mid-size public transit vehicle frequency; last mile connectors to rail or main bus lines). But the industry insists on making misleading (and sometimes outright false) claims of the potential benefits instead of proposing truly compelling benefits.

I welcome thoughtful discussion of what actual benefits robotaxis will provide. To be thoughtful it should take into account negative externalities, system-level outcomes that do not presume 100% personal vehicle use, and viable alternatives that require less investment and less social change.

---

Phil Koopman is a safety advocate who has been doing autonomous vehicle safety for a really long time, and is a supporter of responsible innovation that brings net benefits to all stakeholders.