Evolution of computer methods of dealing with problems is a neat idea, but the amount of time it takes for an evolved AI to be remotely useful is far too long, and it would take many, many years for it to completely realize its own flaws to cover them up.
In order for this research to continue productively, there has to be a way to teach a machine basic fundamentals of what you want it to accomplish without hurting its flexibility to learn in ways we might not think of. It's very difficult.
Good question. I think it has to do a lot with the problem of complexity. Genetic algorithms are very resource demanding, but I think it's a promising area.
I kind of like the idea of using evolution to design tools or solutions. I'm lazy. ;)
Evolution of computer methods of dealing with problems is a neat idea, but the amount of time it takes for an evolved AI to be remotely useful is far too long, and it would take many, many years for it to completely realize its own flaws to cover them up.
In order for this research to continue productively, there has to be a way to teach a machine basic fundamentals of what you want it to accomplish without hurting its flexibility to learn in ways we might not think of. It's very difficult.
cbrhawk1 9 months ago
This has been flagged as spam show
4:47 i'd be so gutted if i was reincarnated as one of those.
tombober 10 months ago
4:47 i'd be so gutted if i was reincarnated as one of those.
tombober 10 months ago
you mean the pokemon golem?
videonfan 2 years ago
I really like this idea of evolving machines on a simulation and then bringing the most successfull ones to the real world.
Very nice.
But will you be able to evolve any useful machine?
StormWolf01 2 years ago
Good question. I think it has to do a lot with the problem of complexity. Genetic algorithms are very resource demanding, but I think it's a promising area.
I kind of like the idea of using evolution to design tools or solutions. I'm lazy. ;)
A3A3adamsan 2 years ago