Or to put it less obliquely, it's completely academic how effective or ineffective an A.I. is, what matters is how much power we surrender to it
Show this thread
Ha ha! This so-called "A.I." machine learning algorithm made a dumb mistake! No fear of them ever taking over the world, right? [the algorithm is put in charge of the world] Ha ha! Hmm
-
-
-
It's super easy to hide behind an algorithm. "Machine says no, so, no. I cannot acknowledge your unique circumstances. Go away. Call the automated line again"
Show this thread -
Note that in the above scenario the system has already "failed" because someone's reached a human. "Ideal" universe, there is only the automated line which always says "no"
Show this thread -
Note that the fact that a machine which always says "no" is not really an "A.I." doesn't affect your ability to hide behind it
Show this thread -
You bought a fridge, now Amazon is recommending more fridges? Ha ha, the system is so stupid! Wait, so what other decisions is this system making? How significant are those decisions? What other systems are those programmers building?
Show this thread -
This thread's a bit of a muddle of ideas so maybe I should stop it here
Show this thread -
Hey let's elect an A.I., it'll be completely unbiased! A small algorithm would be a perfectly fair way to represent the interests of the person who wrote the algorithm
Show this thread
End of conversation
New conversation -
-
-
Pointing out how poorly an AI performs as part of a larger argument against expanding its power is essential.
-
"Well-performing" AIs are also dangerous. In some ways they are significantly more dangerous since people are less likely to challenge/overturn their occasional bad decisions
- 2 more replies
New conversation -
-
-
Thanks. Twitter will use this to make your timeline better. UndoUndo
-
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.