words mean things, and ‘intelligence’ doesn’t mean ‘successful at reproduction’. Indeed, if you look at r and K selection, you will conclude that even across species and genus these are anti-correlated.
-
-
Replying to @djinnius @diachronist and
when we talk about people, intelligence means capacity for symbolic manipulation when we talk about machines, intelligence means capacity for achieving specific ends within an environment, and we call it "general" when those ends look like human desires
1 reply 0 retweets 5 likes -
Replying to @simpolism @djinnius and
the weirdest part of the whole paperclip thing is assuming a "general AI" would even WANT to propagate itself
2 replies 0 retweets 7 likes -
Replying to @woke8yearold @djinnius and
what if the materials it could use to reproduce itself could also be used to create paperclips? it will need to weigh the cost vs benefit of reproduction wrt the goal. the problem is unknown-unknowns: how can it know how effective it will be upon reproducing?
1 reply 0 retweets 1 like -
Replying to @simpolism @woke8yearold and
perhaps the best solution after a point will be for the AI to dismantle itself, turning its own body into paperclips. but all of this is heavily reliant upon the affordances available to such a machine. which is where I have most of my doubts about the entire setup.
1 reply 0 retweets 1 like -
Replying to @simpolism @woke8yearold and
1/3 Thought experiment implies that AI is capable of self-improving to the point of reaching superintelligence. It starts by gaining trust of scientists/programmers to release it from the laboratory, then wins elections on populist platform, then conquers Universe.
1 reply 0 retweets 0 likes -
Replying to @rossiya_vse @simpolism and
2/3 If it just disassembles itself before conquering Universe it'll be a really bad paperclip maximizer and a really stupid superintelligence. We can't know its end goal, but we can safely assume that it will at least do everything to turn all matter into paperclips.
1 reply 0 retweets 0 likes -
Replying to @rossiya_vse @simpolism and
3/3 Its end goal depends on its logic and goals, but if it's just 'Make as many paperclips as possible', one possible outcome is creating an endlessly replicating paperclip-based lifeform in place of our Universe.
1 reply 0 retweets 0 likes
Also, the thought experiment is just an analogy for something that is already most likely happening in real life. Something like urban planning AI suggesting to build a polluting factory in the middle of the city so everybody leaves/dies and reduces traffic.
Loading seems to be taking a while.
Twitter may be over capacity or experiencing a momentary hiccup. Try again or visit Twitter Status for more information.