Will AI-driven Cars Cost Lives or Save Them?
Can AI-Driven Cars Really Save Lives? The Data Reveals a Surprising Answer
Sadly, many people die every year from driving accidents.
But do you know how many?
The World Health Organisation estimates around 1.3 million people die, and over 50 million are injured every year worldwide from road traffic crashes.
It’s an absolute tragedy, as well as the millions of other people, friends, and families that are affected by those who have died.
If that wasn’t bad enough, you see frequent reports in the media of AI autonomous driving cars injuring people in accidents.
Let’s be honest: when you see all this in the news, it can be hard to be optimistic that things can get better, right?
Or, maybe you feel like just tuning out, ignore it, and focus on other things that make you feel good, like watching Juggling Pandas on YouTube.
But I don’t believe you’re the kind of person who doesn’t care about these things and wants to ignore them (but maybe, like me, you do like Juggling Pandas).
I believe you’re the kind of person that does care, and would like things to improve.
So, is there a way you can reduce traffic deaths? Yes, I believe there is.
And this involves learning more about AI-driven cars, not only where they are now, but where they are going soon… and some of this is actually quite surprising…
How safe are Human vs AI drivers?
Human Drivers
Do you know how many people die from traffic accidents each year?
Do you how many are due to human error?
The UK’s Road Risk Assessment estimated that 94% of all car crashes were due to human error.
The top 3 reasons for this human error are:
Failed to look properly (42% of all crashes)
Failing to judge the other person’s path or speed (21% of all crashes)
Careless, reckless or in a hurry (16% of all crashes)
You also know humans are not getting any better or safer at driving.
So you have accepted these millions of people dying or being injured every year by human error, are just normal.
But can we do better? Can we save more lives?
AI Drivers
So are AI autonomous driving cars more or less safer than human drivers right now?
A recent analysis by IDTechEx of Waymo cars, a leader in this space, estimated that their cars were around 0.5% as safe as a human driver.
So no, right now, human drivers are safer an AI drivers.
Are you done? not quite.
We know human drivers are not getting any safer at driving over time, as the IDTechEx report highlights:
Human safety has been fairly stagnant. The rate at which we crash is not changing that much, and further improvement is mostly coming from crash mitigation technology, such as automatic emergency braking systems and blind spot detection
But what about AI drivers?
The same analysis by IDTechEx also looked at whether AI drivers were getting safer over time.
They use ‘Miles per disengagement’ as a proxy measure of driver safety and performance over time.
A chart of the current progress in the safety of AI drivers is shown below.
While the actual average safety of the best 3 AI driver companies in blue has not been increasing as dramatically as predicted, it has still been steadily increasing over time.
Looking at the performance of the best AI driver company Cruise, you can see its safety seems to be doing even better than the prediction.
So what does all this tell us?
This tells us that while AI drivers are not yet as safe as human drivers, unlike human drivers AI driving has been getting better and safer every year.
It’s almost a certainty that at some point, AI drivers will become as safe as human drivers and, then, soon after, even safer and better than human drivers.
This is a crucial fact that most people are unaware of, but you now are.
So, will the public be ready to hand over to AI drivers when they are better than humans?
Pubic Reaction to AI Driving
Given the dramatic increases in the safety of AI autonomous driving cars, the US state of California has been leading the way in trialing these cars and autonomous driving services like Robo taxis.
Indeed, leading AI driver companies like Waymo have been claiming that their AI drivers outperform human drivers for safety.
However, just weeks after granting new licenses, San Francisco has seen protests about the deployment of these AI-driven cars, and this has led the California authorities to scale back, and half the number of AI cars allowed.
We know from the research from IDTechEx that Waymo's claims are perhaps not backed up by evidence, so I have great sympathy for the people currently protesting about the deployment of these AI-driven cars.
I’m not an AI idealist who thinks AI is simply all great, nor am I an AI doomer who believes all AI is terrible.
I am an AI Realist, I believe AI has great potential to improve our lives, and I also think there are some cases where it’s not good to use it.
I would say currently, given that AI drivers are half as safe as average human drivers on average, great care needs to be taken about how AI-driven cars are used in public spaces.
And yet I also recognize and understand, as an AI Consultant, that AI only gets better with more and better data.
I would suggest allowing current AI-driven cars completely free access to all our roads, or banning them completely, are both the wrong approaches.
I believe the right approach is to try to balance allowing gradual exposure of AI-driven cars to real-life public roads and situations as they get better and safer while trying to reduce risk to the public.
But how do you get that balance right between progress and safety?
Regulation, Testing & Education
We must get our laws and regulations right to support this gradual development and transition to the use of AI-driven cars.
You will be pleased to know, that many countries are doing this well on this.
As this TechExplore article recently reported:
More than 50 countries, including Japan, South Korea and the EU member states, have agreed common regulations for vehicles that can take over some driving functions, including having a mandatory black box, the UN announced
What I fear is not being done, is not having enough time, is our education.
In particular, the public understanding of:
Why should you use AI-driven cars?
When is the right time to deploy AI-driven cars to our public spaces?
Why should you use AI-driven cars?
We’re at the paywall. If you want to know what AI means for society with my weekly articles & podcast, with expert analysis of key AI developments and their impact, explained in plain english. Join today for monthly or annual subscription. Hope to help you get smarter about what AI means for society.
Keep reading with a 7-day free trial
Subscribe to The FuturAI to keep reading this post and get 7 days of free access to the full post archives.