Quinn Angstrom
AI thought of the day:
Quinn Angstrom
Sometimes people talk about the idea of Strong AI coming forth unintentionally through emergent programming experiments
Quinn Angstrom
Here's my thought on that
Quinn Angstrom
when people write AI to 'evolve' solutions for specific problems
Quinn Angstrom
it's kind of like a more directed version of a million monkeys with typewriters trying to write all the works of Shakespeare.
Quinn Angstrom
For that kind of program to unintentionally create a general intelligence
Quinn Angstrom
would be like one of the monkeys accidentally using his typewriter to type up Shakespeare.
Quinn Angstrom
The man.
Quinn Angstrom
Just accidentally pressing keys on the typewriter in such a way that it creates life.
Vehrec
That does seem extra-unlikely. Even the DNA of Shakespeare is pretty unlikely. Probably moreso than his complete works.
Quinn Angstrom
yeah, to be clear, I don't mean it types up all the information needed to reproduce Shakespeare.
Quinn Angstrom
I mean actually creating a living, functional human being.
Vehrec
Obviously, people worry about this because evolutionary processes have produced one example of general purpose intelligence (i.e. humans). But the kind of evolution in question is very different.
Quinn Angstrom
the tools we've created are fundamentally unsuited to the task of creating strong AI, intentionally or otherwise.
Vehrec
Yeah, I won't doubt you on that.
Villain
AI fear annoys me for this reason
Villain
I couldn't have articulated it quite so well, not being a computer expert, but the odds of it just seem so infinitesimally small
TemporalWolf
Yeah, same here. Thanks for articulating it far better than I could.
Quinn Angstrom
it's so unlikely that the word 'unlikely' doesn't make sense for it.
Quinn Angstrom
It's like a tornado striking a junkyard and coincidentally assembling a cold fusion reactor.
Vehrec
Even creating a single AI program that can play both chess and checkers seems like something that would not be virtually impossible using these methods.
magnet
Strong AI is definitely not going to happen accidentally
Vehrec
That would be virtually impossible, gaaaah, type your words properly!
magnet
it is totally going to be made on purpose eventually, though, which is why the value alignment problem is important
載入新的回覆