​The Stig Is Terrified Of Self-Driving Cars

Illustration for article titled ​The Stig Is Terrified Of Self-Driving Cars

It's easy to understand why some of us don't like the idea of autonomous cars. We enjoy the drive, adore the freedom, and practiced heel-toe downshifts for too damn long to give it up. But Ben Collins – former Stig and current halfwit – apparently gets paid to play fear mongering pundit rather than rational commentator.

"I think it's terrifying," is how Collins sets up the softballs. "Robots are fantastic as dishwashers, but they don't make very good drivers."

Agreed! Our magical robotic dishwashers are amazing! And ironically, they're treated the same as cars by people who consider both to be appliances for doing menial shit.


Collins then goes on to explain that computers aren't capable of decision making and that all the necessary systems aren't up to snuff. (Exactly why we threw out the first dishwashers and continued to scrub pots and pans by hand. Technology never improves!)

One place we do agree is that drivers should be better trained. It's just too bad that both the public and politicians aren't willing to step up, which is one of the reasons autonomous technology is being developed.

You can watch the entire segment here, because he does have a few honest critiques, but most of it comes off as knee jerk, faux-rage rather than reasoned argument.

Share This Story

Get our `newsletter`


KusabiSensei - Captain of the Toronto Maple Leafs

Putting a computer in control of a car, easy. Robotics has solved that for some time.

Getting a computer to be intelligent enough (or, if you want to be pedantic, to be able to fake intelligence enough) to deal with humans and their non-rational decisions in other cars? That's still not quite there yet.

Setting up a course of parameters and having computers drive within that is fine, since it's systematically possible to determine what the other robots will do.

But remember: Humans can do one thing very well that classic computers cannot (I will disregard quantum computers for the time being, but they may very well solve this problem)

They can learn. Much quicker than a computer can machine learn (If you don't believe me, watch Watson play Jeopardy with Ken Jennings and Brad Rutter and watch for Watson's mistakes). Humans can also use intelligent prediction to make much more educated guesses with an innate confidence level, which is still not simple to do without a supercomputer.

Just getting a computer to have a conversation well enough to fool someone into thinking they are talking to a real person is hard enough. Adding control of a car of that level of reasoning, analysis, and response is not simple at present.

Not saying it won't get there, but right now, I wouldn't want to have a computer in control of a car outside of a closed course without a manual override.