You probably remember those pesky website tests to verify you’re human using CAPTCHA. Initially they were created with character recognition in mind, so you had to type all kinds of words, then combination of words and gibberish.

That is a fine way to tell the difference between a non character recognition system, such as machines from humans. Trouble is that this wasn’t only a test, but a training data goldmine for computers. What happened next? That’s right, the algorithms “learned” to recognize what tells us apart from them, thus beating us at our own game.

Then came image recognition – you had to tell which squares contained busses etc. Alas, same story. We helped train the system.

Then came “I’m not a robot”, which identified movement patterns. Doesn’t tell a genius to, right?

If you’ve tried using TikTok for a while, it would now present you with an image you’d have to align. It used to be a linear slider. Today it’s a rotating puzzle. I’m guessing AI caught up with that, too?

Here’s the thing: one of tech’s side effects is providing solutions to problems tech created in the first place. You’d guess it’s a bad thing and you’d be wrong, because perpetual problems are perpetual business opportunities.

And it’s also a cat and mouse chase. For every new lock there’s going to be a new lock pick. It’s because locks are technical and stealing is human. In a place without a human incentive for crime, there’s hardly a reason to lock the doors. I know of times and places where this existed.

This is a toxic and wasteful combination. The solution providers aren’t worried by their shortcomings. Running out of printer ink is in their interest, not yours. We just have to keep on working to keep up with security, only to be presented with a new scheme later. The only comfort is although we are working in corporate best interests, the cost isn’t that awful yet.

Worth a watch: