Reality Check on Autonomous Cars

Not a week goes by without someone extolling the imminent virtues of autonomous vehicles. It’s gotten to the point where some see them solving pretty much everything. Which, obviously, they’re not going to do – so let’s take a quick look of what impacts we can really expect, and when.


What, by the way, is a train image doing on a post about autonomous cars? It’s because trains already provide the user experience that people are likely to most rave about when they get their autonomous car. We’ll return to this later.

Some, like RAND, provide good, balanced reports on autonomous cars. Others, like Accenture, focus on the eccentric (“Autonomous vehicles can expand consumers’ access to banking, using the car to pay for fuel and tolls” – wot?! :)) while yet others go for the utopia-model: this Investopedia article is a good example: autonomous vehicles will supposedly “dramatically reduce the number of cars” and make for “greener urban areas“, among other impacts.

Let me begin by saying I love the concept of autonomous cars, and I believe they will be a net benefit to society and will change a number of things – some dramatically, some less so.

Will they reduce vehicle ownership? Maybe, but it’ll be a reduction, not elimination, and the shift is likely to be generational in timeframe. While fewer young people now own a car, those predicting significant shifts in attitudes to ownership ought to keep in mind humans aren’t exactly logical creatures, and ownership of “stuff” is rarely driven by rational reasons alone. I can also see the reverse trend being possible, such as affluent people buying cars for their kids – kids too young to drive themselves – to get around with.

Will they reduce accidents? Very likely yes, and this is the single biggest human benefit of them. I fully expect that in another 50 years, driving yourself on normal roads is likely to be either illegal or ridiculously expensive because of insurance.

Will they reduce emissions? No. Of course, their driving style might be more efficient, but that alone is not a huge impact – especially when you factor in the increased miles driven (see below). Even if we assume they’ll all be electric, that won’t automatically have a positive impact – because depending on the way your electricity is produced, EVs can pollute more than traditional combustion engine-equipped cars. This is the case here in Victoria for example. Thanks, coal.

Will they reduce miles driven? Almost certainly not. This is even contradictory to the other often expected impact of reducing vehicle ownership.

How so? Barring substantial changes in human behaviour (e.g. dramatically reduced mobility), the only way to reduce the number of vehicles is to use them more efficiently – share them. Given their (then autonomous) movement from one passenger in need of them to the next, they will actually increase miles driven – and as such, also increase emissions, and congestion.

Even if we assume a more conservative “no sharing” deployment of autonomous cars, they’ll be driven more and will worsen congestion. Imagine: you drive – well, are driven – to work. Do you tell the car then to park in the busy CBD area at a cost of, say, $50 per day – or tell it to go find a cheaper spot further away? As long as it’s there to pick you up when you get off work, chances are you’ll send it somewhere cheaper. Or when you need to stop somewhere for 10 minutes and there are no parking spots? Just tell the car to drive around until it’s time to pick you up.

Not to mention that when driving becomes more convenient, people are likely to do it more. For example: why fly from Melbourne to Sydney, when you can just head out in the evening in your comfortable, self-driving vehicle that provides a lie-flat bed for you to sleep in? On the less extreme end of the scale, they will likely make longer commutes more feasible – again leading to more miles driven.

Will they eliminate traffic jams and ease congestion? No, as above. But they will make traffic jams more bearable so I guess that’s something.

What about the user experience; what will that be like? Calling for your car like KITT was summoned in Knight Rider will be neat alright, but the most meaningful impact is elsewhere: we all know driving in traffic jams, in technical terms, sucks. It’s stressful, and for some, road rage ensues. Having an autonomous car will improve that experience significantly – it’ll reduce stress as the “driver” can concentrate on other things: relaxing, working, reading, even sleeping.

Which brings us back to the trains. I suspect something along the lines of “my commutes are SO much better with my self-driving car” will be the most common thing the owners of autonomous vehicles will be raving about.

I find that rather ironic, because that “hey I can read a book or sleep during my commute!“-experience is exactly what users of public transport, where available and done right, have enjoyed for decades. Everything old will be new again 🙂

Over time, Level 4 autonomous vehicles will allow a complete re-design of the car interior which has already led to some exciting concepts. But again: if you make car travel something really enjoyable, it’s likely to lead more of it being done.


But most of all, none of this will happen overnight. The current commercially sold state-of-the-art vehicles are Teslas, which are “only” Level 3 autonomous vehicles. Level 3, while providing added convenience and safety, allows for none of the significant societal changes to take place yet – cars aren’t really truly autonomous until they reach Level 4, which is still some way off.

A number of manufacturers have stated they will have autonomous models on the market by 2020. Chances are the majority will be Level 3 autonomous. Taking an extremely optimistic assumption, let’s say that Level 4 autonomous capabilities – for all roads and all conditions – are perfected by 2025 (which I think is overly optimistic), and that they will be in every single vehicle sold in 2030.

There is little reason to believe that vehicle replacement cycles have the potential to dramatically quicken – and there would be production bottlenecks to overcome, too. That means that by 2040, “only” approximately 50% of the vehicles will be autonomous and we’d approach 90% penetration only by 2050. Some benefits of autonomous cars, such as increasing road capacity (through smaller distances between cars and potentially higher speeds) won’t reach full potential until the penetration is dominant.

If 2050 seems like “too long”, consider that the average age of vehicles is about 10 years, give or take a couple years (11.5yrs in the USA). And there’s a long tail – did you know there are 14 million vehicles over 25 years old on the road in the USA, and a total of 58 million cars over 16 years old? (58 million is over 20% of the light vehicle “install base” of 258M in the US).

2050 doesn’t sound like an overnight revolution to utopia, now does it? And remember, this is under extremely optimistic assumptions.

In other words, remember Amara’s Law:

We tend to overestimate the effect of a technology in the short run and underestimate the effect in the long run.

While realism doesn’t sell papers, urban planners, take note: autonomous cars will not solve your infrastructure woes. Neither will Uber or taxi drivers all become imminently unemployed – although eventually they will, so I wouldn’t recommend it as a long-term career goal for kids.

The bottom line? As fast-moving and exciting as the autonomous cars thing is, it won’t happen overnight – and the impacts are not so clear-cut-positive as many would have you believe.

This entry was posted in Culture, Technology and tagged , , , . Bookmark the permalink.

Leave a Reply

Your email address will not be published. Required fields are marked *