Phani Raj Kuchibhotla

Curious Curator

Direction for AGI

If not immediately , we will eventually achieve Artificial General Intelligence, where machines are much smarter and can perform any task a human can perform and sometimes better than humans. But do we want to steer in that direction and what precautions do we want to take before we reach there? I found this interesting video today, where Max Tegmark talks about the same.

He says, we invented fire, burnt our hands and only after a few centuries we invented fire extinguisher. Even with AI it’s going to be the same, but with AGI we may not get a second chance to invent a way to mitigate the influence/impact of AGI. So, this time we have to more careful what we are inventing with AI and make sure we have the principles and ethics in place such that AGI doesn’t overpower but rather empower us.

We are now at the best place in the history or human evolution to make sure we follow enough safety precautions and think through kind of society we want to build with AGI. Build a feature that is thought thru enough as to how we live with machines together and how we complement each other.

Sometime I wonder, once AGI is achieved, what are humans going to do all day? Any thing you wanted to do, a machine can do the same in better, efficient and cheaper way. So will humans be reduced to just reproduction?

Will AI get so much aware that they are going to decide what humans do and don’t?

So it’s in the hands of AI scientists now to consult the world and define the guidelines as to what kind of future we want to live in? It’s not going to be easy to think thru decades into the future and decide what we want, but can always say what we don’t want to happen.

Here is a TED talk, that summarizes the problems we are about to face with AGI

Making superintelligent A.I. is a really hard challenge. Making superintelligent A.I. that is safe involves some additional challenge on top of that. The risk is that if somebody figures out how to crack the first challenge without also having cracked the additional challenge of ensuring perfect safety.

…….we should work out a solution to the control problem in advance, so that we have it available by the time it is needed.

If we dont plan the direction from now onwards, it may be too late.