Categories
Process Control

Happy new year, and some thoughts on AI

I’m Jason Firth.

A lot has been said about AI of late, the idea that AI could “evolve” and become a true intelligence.

With respect to any system I’ve seen so far, that sounds plausible, except that it isn’t. Much like a stuffed doll will never evolve into a human because the fundamental stuff that makes up both are completely different, “Artificial Intelligence” and actual intelligence are made up of things that are absolutely different.  Take the blinker on your car. Is there a chance it will ever become truly self-aware? The answer is clearly “no”. It is a switch and a timer circuit and a couple light bulbs, designed by a human being to do one and exactly one thing. How about if you replace the switch with a voice activation? It’s still a switch and a timer and a couple light bulbs. How about if you have a wave file play saying “Activating turn Signal” when you turn it on? You’ve made the interface more human compatible, but the fundamental stuff that makes it up is the same. It’s still a timer circuit at its root. The logic behind these three “Artificial intelligences” are very similar. Every single AI I’ve ever seen is a more complicated version of the exact same concept as the turn signal. It’s a purpose built machine made for excelling at one task. Deep Blue can beat Garry Kasparov a thousand times at Chess, but it will never beat him at Mario Kart unless a human intervenes and effectively creates a new device for beating Mario Kart. It will never beat him at writing a song unless a human intervenes and creates a new device for writing songs. It will never beat him at writing poems unless a human intervenes and creates a new device for writing poems. By contrast, the human mind invented chess, and Mario Kart, and songs and poetry, then determined what was a good state and a bad state, and then determined methods to get to the good state. The human mind wired itself to do this, along with a thousand other things that were required to get there. No human ever opened up someone’s brain to wire up a chess player or a mario kart player or a musician or a poet. As long as that distinction exists, artificial intelligence will always actually be a mere tool created by a human intelligence. Such intelligence “evolves” by the humans getting smarter and applying different algorithms to different problems successfully, not by solving problems itself. The day that an AI identifies and quantifies a problem then comes up with a solution on its own, that’s the day I’d be concerned about a true intelligence developing. Until then, it’s purely science fiction. Thanks for reading!

Leave a Reply

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