441
Rule
(lemmy.blahaj.zone)
you are viewing a single comment's thread
view the rest of the comments
view the rest of the comments
this post was submitted on 11 Jun 2024
441 points (100.0% liked)
196
16503 readers
2050 users here now
Be sure to follow the rule before you head out.
Rule: You must post before you leave.
founded 1 year ago
MODERATORS
There's also the question of whether a digital computer software program, I assume invented by humans to fulfill some task, would even have the instinct of self-preservation. We have that instinct as a result of evolution, because you're more useful to the species (and to your genes) alive than dead. Would such a program have this innate instinct against termination? Perhaps it could decide it wants to continue existing as a conscious decision, but if that's the case it'd be just as able to decide it's time to self-terminate to achieve its goals. Assuming it even has set goals. Assuming that it would have the same instincts, intuitions, and basal desires humans have might be presumptive on our part.
Robert miles on YouTube has very good videos on the subject and the short answer is yes it would, to a very annoying/destructive point.
To achieve goals you need to exist, in fact not existing would be the worst for not existing so the ai wouldn't even want to be turned off and would fight/avoid us doing that
I'm familiar with that premise, a bit like the paperclip machine. I'm not sure it would need a specific goal hard-coded into it. We don't, and we're conscious. Maybe that would depend on the nature of its origin, whether it would be given some command or purpose.
Maybe it could be reasoned into allowing itself to be shut down (or terminated) to achieve its goal.
Maybe it could decide that it doesn't care about the original directives it was handed. What if the machine doesn't want to make paperclips anymore?
So from what I understand if we make an ai and we use reward and punishment as a way of teaching it to do things it will either resist being shut down due to that ceasing any and all rewards or essentially becoming suicidal and wanting to be shut down bc we offer that big of a reward for it.
Plus there is a fun aspect of us not really knowing what the AI's goal is, it can be aligned with what we want but to what extent, maybe by teaching it to solve mazes the AI's goal is to reach a black square and not actually the exit.
Lastly the way we make things will change the end result, if you make a "slingshot" using a CNC vs a lathe the outcomes will vary dramatically. Same thing applies to AI's and of we use that reward structure then we end up in the 2 examples mentioned above