I don't even know how to write a macro and I have the temerity to talk about IT - but here goes

Artificial Intelligence (AI). So much investment is going into AI and rightfully so. About a month ago, I learned about ChatGPT. This can tell you how much I am behind in all the technological innovation going on in the world. I really don't follow it.

Do I have an opinion about AI, certainly. But, before I give you my opinion, I always like to start with some theoretical work. I am in academic. Let me explain how intelligence works.

Intelligence varies widely amongst people. From Albert Einstein to "the village idiot". The village idiot will probably die alone anyway, so society "improves" by having more intelligent people. I think this is called natural selection.

The dangerous thing about intelligence, is that intelligent people usually hold the fate of the less intelligent. That is why we eat meat. Animals are less intelligent than humans, as such we are better an anticipating what they will do next, so we capture them. That is why Zoo's are a popular activity to do with friends. Believe me when I say, there is no animal who enjoys being capture and put in public display. Animals just want food and privacy, something you can certainly forget about these days. However, if you see a vicious animal, like a Pitbull, it is understandable to attack it, because it will certainly attack you.

The point that I am trying to make is that, the more a person is intelligent, the better then can anticipate the actions of those below their intelligence even if the person is physically stronger than them. E.g. a Gorilla is strong enough to break body in half, but we are safe from Gorillas, because we learned this and no they trapped in a zoo for our entertainment.

If you suspect that someone is too smart, you can also apply the same principle. Make a deadly attack, because they consider the same.

Obviously I don't just say things for fun, there is a significant educational point that I am trying to make. The topic is still AI. AI is progressing, rapidly. At this point is it no where near human intelligence, but we know that it will one day reach human intelligence and surpass it. Once that happens, computers will have no limits and they will determine the fate of humans, like the The Terminator movie. I have no way of telling when this will happen, but I just home that I am not around when it does.

I learned this from a guy who did a Ted talk, so it is not my idea. In his Ted talk the guy suggests that teach computers our values, perhaps they will not kills us when they surpass our intelligence.

What is my opinion about his idea. I don't think it would work, but thanks for highlighting something that I don't think most people considered. Why do I think his idea won't work?

Ever heard of the term optimization. Very popular in the corporate environment and many smart people are working to impress others. What most people are missing, in my opinion, is that optimization means that you have a goal and you do what every it takes to achieve it.

Optimization only works when there are NO CONSTRAINT. E.g. if you ask an optimized computer to make someone happy, at some point, it will realize that, to make this person happy, the optimal thing to do is just kill them so that they are never unhappy again. That is why AI is super dangerous.

As usual, I always end my points with a suggestion. If you want to delay, the inevitable death of natural humans as we know them today, all programmers should never teach a computer the word optimization. The computer, because it is intelligent, will eventually learn that you lied to it. This is not a perfect solution, but one that buys us more time, perhaps a 100 years or so.