AI is getting there, apparently, but do we really even want to go there?

AI is
Image by 1STunningArt | Bigstockphoto

AI is ‘getting there’ according to almost anyone you care to ask. It is at the point where politicians are expounding on an ‘AI driven future, the saviour of the economy’ – without really having a clue what they are talking about.

AI is indeed getting there, in some areas. We hear about experiments where AI programs are set tasks and come up with unlikely but highly effective solutions. Then AI sets the next, harder, task – and solve that too.

But there is too much fuzzy, hype-driven, thinking going on.

We are, according to our new favourite hype model, at the peak of Mt Stupid.

Some will say that AI cannot be intelligent because it doesn’t ‘think’ like a human brain. Some will spend their entire professional lives trying to understand the brain (which is good) and then trying to invent a process that mimics the brain. That is probably less good and astonishingly ambitious.

AI works in environments with boundaries and ineffective in environments where the rules are changing – or where there are suddenly no rules. We cannot teach AI each and every permutation to the rules because we do not know them. Confronted by a situation where we (humans) come across something we have never encountered before we use common sense, intuition, instinct.

Can we ever teach AI common sense or instinct?

Almost certainly not and if we do eventually figure out how to get close, it must, surely, fall a long way short of real human instinct.

AI is also only as good as the training and data we provide and if we asked AI to pick the very worst entity to provide that training, it would pick humans, without a doubt.

We are prejudiced, we are rude, our rules change every day, our opinions are based on what the most popular people are saying (who probably know as little as we do). We are flawed.

And we are going to train AI from that foundation?

AI will only be as good or evil as we are and we are very good, very evil and everything in between.

AI is also more effective than we are in implementing decisions quickly and clinically.

So, we are investing countless time and energy into something that is as flawed, addictive, good and evil as we are.

If we achieve that, goodness only knows what we will have created and what decisions AI will make about us, the planet and, well, everything.

Related article:

AI might have an addictive personality and that is not good

Be the first to comment

What do you think?

This site uses Akismet to reduce spam. Learn how your comment data is processed.