Conversation
Notices
-
"nobody is out to get you. you are not important enough for the powers that be to spend their time and attention at"
>technology that allows for effortless, automated and relentless personalized psychological warfare is invented and widespread
uh is it justified to be paranoid yet
-
@augustus
>how to cut through the noise? seems an impossible task
it's not impossible at all if it doesn't need to be acted upon by a human. if a shitty company like youtube can identify a second worth of copyrighted material playing in the background of some dudes 12 views hawaii vacation vlog...
-
@augustus it's not an == operation. it's a neural network pattern matching one that's almost as good as a billion wagecucks at picking out stuff from the petabytes of shit that gets uploaded every day.
>. how do you analyze posts and behaviour and compare them to a likelihood function of being a criminal
same thing. when your data set is wide enough you can pattern match for vars that humans haven't even made up names for yet.
-
@augustus wow who would have thought introducing force multipliers and reducing the relative the value of the individual would result in an insurmountable physical and computational power gap between the rulers and the ruled.
professor ted.jpg
-
@augustus oops! accidentally found a way to automate the entire clerical class giving the average megacorp of 2019 the power of a hundred billion human accountants, lawyers and private investigators working full time non stop 24/7 with little need for oversight! i
-
@augustus @mono
matching those video samples and matching two text samples are the same problem, and so both doable, and regularly done, now
the problem is that these "ais" have no intuition. they can become reasonably to very accurate, depending on the training set, but have no way to "step back" and see when they're doing something stupid in those 1-10% of bad matches. they're just not-especially-complex-in-principle-but-filled-with-a-mountain-of-fiddly-details-in-practice black boxes that estimate probabilities. brute-forcing things with a mountain of if-then conditions
and so things like these pictures happen
everyone's using them anyways, though, because they "mostly" work, meaning people get caught by those false positives and penalised for nothing, seemingly at random, and by not-actually-entities who can't be talked out of doing stupid things
waifu8.png
waifu13.png
-
@augustus just think of a list of all the things an individual is on close-to-equal in terms of ability is with society.
In 1999, Kasparov played a game of chess against "the world" (actually a consortium of the other best chess players in the world) and won, because having lots of people doesn't actually make you any smarter past a certain point.
In 2019, computer chess is basically a solved problem. Alphazero plays chess on a level several magnitudes higher than Carlsen can. You can not beat a chess computer without another chess computer.
The more problems like this we solve, the less and less options an individual has to fight back against a thought reading godlike superweapon owned by assholes who sexually get off to debasing you.
-
@shmibs @augustus
>the problem is that these "ais" have no intuition. they can become reasonably to very accurate, depending on the training set, but have no way to "step back" and see when they're doing something stupid in those 1-10% of bad matches
this will be solved to an acceptable rate within a couple years of development, especially as automated data cleaning and interpretation improves.
-
@shmibs @augustus also stylegan is a bad example because it's as primitive as it gets when it comes to self awareness and result checking.
-
@mono @augustus
it is a fundamental problem in the current systems, though. all that's really changing is twiddling a few knobs and throwing bigger datasets at things to try to drive down uncertainty as far as possible. but some things just inherently have small datasets
-
@augustus
>because they're trained on datasets of the real best human players move sequences
alphazero had no human input at all. it taught itself how to play chess by competing with itself. it won over the next smartest AI 78 vs 0.
they took the same code and then applied it to go, and within a couple months of self training, it beat a grandmaster.
In Naraka (permanently)
バツ子（喉痛いよX_X
All senooken JP Social content and data are available under the 
