Artificial Intelligence, your car, and you – Part two

You are good at this. Very good indeed. If you live in the US, you will statistically drive 74 million miles, or over 5000 years behind the wheel before you have a fatal accident. And this is all based on very rudimentary sensor or perception hardware. You have just two little optical cameras, mounted about 50mm apart, in the front of your skull. And they are not even very good cameras – they have no ability to see infrared, for example, and they may even need corrective lenses. You have no lidars, and not even a little radar mounted in your forehead. And yet you manage to drive, on the whole, remarkably safely. And you pull off this feat because those deep neural nets in your brain are massively impressive at pattern recognition and interpretation.

But AIs are even more impressive – and they are getting faster and better at a rate that far outstrips the slow drip-feed of biological evolution. When we give them much better sensor hardware than you have – say, a 360-degree suite of multiple radars, cameras and lidars – that computer will (not might, will) surpass you at the task of driving.

But hang on, I hear you say. We’ve heard this stuff before. Autonomous vehicles have been talked about forever. Aren’t they just another Silicon Valley investment/hype bubble that will never see the light of day? And didn’t I read just recently that one of those Cruise things knocked a guy over in San Francisco? Surely all this is just a pipe dream. Autonomous vehicles don’t actually exist, do they?

They do. Autonomous vehicles drove almost six million miles in California alone in 2023 – up from four million miles in 2022 and about two million in 2021. The mathematics fiends among you will have spotted that this is an exponentially increasing  curve – and we all know what happens to things that go exponential. You can hail a fully autonomous ‘taxi’ right now in Arizona, San Francisco, Seoul in South Korea and several major cities in China. And trust me, seeing is believing. It’s one thing to ride in an ‘autonomous demonstrator’ on a test track with a safety driver hovering their hands over the wheel, foot poised over the brake pedal, just in case. It’s another thing altogether to ride in the back, with no other human on board as an AV pulls out into real traffic on a public road and takes you safely to the destination, steering wheel spinning smoothly in the air as if a highly competent but invisible pair of hands guide it.

And autonomous, shared cars really will – in time – be a revolution, not a mere evolution. They will make ‘driving’ (or getting around in cars) in cities a safer, cheaper and less dreary task than it is today. They will help solve the ridiculous situation we’ve created today, where 87 per cent of journeys are made in a two-tonne, five-seat car with just one person on board. They will also help us resolve another terrifying statistic – the fact that cars spend 95 per cent of their useful life stationary. You read that right – cars spend only 5 per cent of their life doing what they were actually designed to do: move. No other large capital asset has so poor a usage rate. And as we approach 10 billion souls on this ship we call a planet, these statistics are no longer sustainable. Autonomous cars, which will be owned and operated by fleets rather than individuals, will help move these numbers to something a bit more, well, reasonable.

But it’s not all sweetness and robotic light. When autonomous vehicles scale up and move beyond Silicon Valley and China, it is inevitable that fewer people will buy ‘private’ cars. They will make less and less sense to the increasing global population that will live in large cities. Improved public transport and shared autonomous vehicles will start to supplant the privately owned car. Which means a thing marketing folks call TIV, or Total Industry Volume, will inevitably fall. In short, the number of cars built and sold in the world will start to decline from today’s figure of around 67 million passenger cars per year to something significantly smaller.

Now, this is a terrifying prospect for the traditional car industry. For over a hundred years, it has thrived on building a large chunk of metal and plastic, selling it to you and I for cash-money, and betting that we will come back every three or so years for another. Autonomous vehicles don’t represent a gentle swell on the ocean of the car business as we know it – they represent a violent sea-change. And every OEM on the planet is right now spending a lot of brain-power, stress and treasure on figuring out how to position themselves in the coming storm. It will have a profound impact on the industry – but I don’t pretend to have the answers on what exactly those changes will look like. But make no mistake – AI in the form of autonomous technology is to today’s car industry what the Model T was to the farmer’s horse, or what digital photography was to a nice stable film business called Kodak.

But to end on a very selfish note, what about us? As enthusiasts not just of cars but of driving cars, need we be worried about autonomous vehicles? Not at all. Huge stretches of the world’s planet will not have the density of population to sustain the autonomous-shared business model. Nor will governments legislate hastily to ban human beings from our roads. Even when it is demonstrably safer to allow the machine to drive, can you imagine the political courage it would require for any political party to try to get us human drivers out from behind the wheel?

So we can look to a future where we no longer have to drive, but we can if we so choose. We will no longer have to sit slumped and frustrated behind the steering wheel in horrible wet Friday-night traffic jams, crawling listlessly through London or Shanghai. But I guarantee you that a human-driven 911 will be ripping up a canyon somewhere in California, or over a mountain pass in the Alps.

AI? Bring ’em on. But some patience is required. The revolution is coming, but not just yet.