The AI apocalypse and why it will (probably) happen

This is my fourth or fifth article in the matter and it seems that this streak will not end soon.

Almost everyone I know, talks about AGI, and how computers will get out of control, they will understand everything and we are doomed to be exterminated.

The fallacy

The main fallacy around this matter, is that in order to be destroyed, we need to achieve AGI first or create also artificial consciousness. Nothing of this will actually happen, only the apocalypse.

Why it will happen? Simple, we do not need to achieve AGI to make AI dangerous.

War games

Do you remember the movie War Games (1983)?

In that movie, the protagonist, accidentally dial-up into a military mainframe, and begin to play a game of global conflict. The computer is confused and is ready to begin War World 3, by launching nukes.

How far we are right now from this reality? Back then I had a ZX Spectrum 48K+ and as you can imagine, training LLMs with that Z80, what really out of the question. :)

But now, 41 years later, we are there. Someone, with a modest desktop can built an agent with that kind of capabilities and the bad news is that it is not really difficult to do it. It is easy.

It is a matter of time actually, for someone to create some kind of weird prototype or military project.

Lack of morals

So, the problem will be that computers will now how to do stuff, they will make decisions, but they will only have probabilities to pick the best decisions. Right now, we are trying somehow to approximate intelligence.

Consciousness and morality are far away, nobody has a clue how to approach or model them. Without them, AI is unable to isolate good from evil.

And probably they wont.

So, what now?

Nothing, actually. We are all sitting comfortably and waiting for robots to become useful. The one will lead to the next. Just grab your pop-corns and we will see what will happen in the end.