Putting the A into AI

In a recent episode of Top Gear (S22E06), the hosts got going about driverless cars and how one day these cars are going to being making decisions about whether they should kill a bunch of kids or let you die instead. And it’s a funny little skit they do, but many people seem to believe that this is a current issue.

What I find surprising, is that no one is actually worried about these things, until someone starts a conversation about it, and then it’s fun to joke about, but in a serious manner. It’s the same thing we saw in the movie adaptation of iRobot, where Will Smith hates robots because one chose to save his life over a child’s, because the odds were better. It’s also why during the rest of the movie the robots try to imprison us, for the ‘better good’.

But it’s called Artificial Intelligence for a reason, mainly because it’s not real. It’s true that we’re doubling the processing capabilities of computers on a very frequent basis, but we’re still decades away from being able to get a computer to take in and process enough data to be able to make that kind of decision, and then it’s still only making a decision based on what we’ve taught it, there is no intelligence in the true sense of the word. Nevermind who gets to make those decisions anyway? Is there an ISO standard which dictates the hierarchy of people. For example: two 20 year olds > one 50 year old? No. Not any time soon anyway. And I don’t know if I want my car making these kinds of decisions in any case. I don’t want it to know that much.

The sad (maybe) truth is that it’s really really hard for a computer to quickly recognise different kinds of objects. It can quite possibly recognise obstructions, but to differentiate what these all are in any usable time frame is nigh impossible. More likely the car will see it’s about to crash, will slam on brakes and determine if there is an area to its left (for RHD cars) that it can swerve and avoid hitting something else. This is where we are currently. And it’s great, because as much as we as people can analyse, people make bad decisions. More often than not though, the car will avoid being in situations where it has to make this kind of decision. It won’t be tailgaiting like humans do, it’ll have a nice following distance that gives it time to react. And it can react quicker than us.

I’ve written about autonomous vehicles before, the ‘Yay’, but ‘ahh’ feelings I have towards them. The biggest factor against them at the moment is the lack of regulations governing them. And no one really knows how to react to this. There are too many uncontrolled variables, which I don’t see being solved in the next 10 years, not on public roads in any case.

Some of my posts (such as this one), I’ll have the idea for the post, sketch out one or two paragraphs, but then it sits as a draft for almost 3 months. And as time goes on, more and more people write and speak about the topic, such as the below 2 articles which have al come out in this time :) It is an exciting time to be in, and I look forward to seeing how the rate of penetration increases. I’m not as optimistic as others may be, but agree that it is the future, and hope to be able to be a part of it.

Elon Musk mentioned it in a recent key note address

Another very relevant link: Self-driving cars and the Trolley problem

Leave a Reply

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