|
https://twitter.com/billyperrigo/status/1615682180201447425 https://twitter.com/billyperrigo/status/1615682186409041920 https://twitter.com/billyperrigo/status/1615682192113270785 https://twitter.com/billyperrigo/status/1615682196878000130 @computer_genious.eth: I believe that AI is an autonomous being that can one day learn to feel emotions Some dude named Abuya: It's my job to tell the it to stop constantly posting nazi snuff porn. I wish I got healthcare benefits
|
# ? Jan 18, 2023 16:22 |
|
|
# ? May 6, 2024 05:57 |
|
eSports Chaebol posted:ATTENTION! ATTENTION PLEASE! thank you for your attentioncoins now read this you are forgetting about Eripsa eripsa was right
|
# ? Jan 18, 2023 16:23 |
|
bvj191jgl7bBsqF5m posted:https://twitter.com/billyperrigo/status/1615682180201447425 Ah yes, the old Mechanical Turk gambit with a neoliberal twist
|
# ? Jan 18, 2023 16:28 |
|
bvj191jgl7bBsqF5m posted:https://twitter.com/billyperrigo/status/1615682180201447425 So the AI does have a soul
|
# ? Jan 18, 2023 16:36 |
|
Gumball Gumption posted:So the AI does have a soul Yes, the soul of Henry Ford! https://twitter.com/ZaneGTCooper/status/1615577714836275200
|
# ? Jan 18, 2023 16:57 |
|
https://twitter.com/ZaneGTCooper/status/1615586981177708545 creating chatbot versions of famous historical figures just to make them repeatedly apologize for and disown the very public views and beliefs of the actual people is deeply strange. Can't wait for Hitler Chatbot to tell us that he deeply regrets the the writings contained in Mein Kampf and while they were shaped by the intellectual and cultural climate of the time, they don't reflect his true beliefs.
|
# ? Jan 18, 2023 17:02 |
|
Consider that "humans actively editing out the ai's ability to say explicitly horrible poo poo" and "chatbots writing apologia" might be connected things.
|
# ? Jan 18, 2023 17:06 |
|
anyway, i think so far with neural computation, we have only managed to replicate individual neural systems, or create pure facsimile systems capable of following orders and synthetizing reasonable answers to queries, so long as the questions are parsed by the user to fit the training data available to the AI. and that training data isn't composed of 4chan posts. and all this is done without any of the ambivalence (of aim) in decision that accompanies actual animal behaviour, as the AI is simply seeking to optimize a number, representing how good an answer is, to be as high as possible. this is fine for systems meant to do one thing, and one thing only, to fulfil a task in the same sense that the frontal part of the visual cortex fulfills the task of intreting optical nerve signals, and yet human intelligence has very little to do with the presence of optical nerve signals themselves, or even the presence of the visual cortex itself. in congenitally blind individuals, damage (or interference by fMRI) to the visual cortex will impair things like braille reading, mathematics with effect growing function to complexity, and ability to navigate spaces interestingly enough. im pretty high so im going to stop here
|
# ? Jan 18, 2023 17:08 |
|
|
# ? Jan 18, 2023 17:09 |
|
Snake! Snaaaaake! SNAAAAAKE!!!
|
# ? Jan 18, 2023 17:14 |
|
KirbyKhan posted:Snake! Snaaaaake! SNAAAAAKE!!! lmfao
|
# ? Jan 18, 2023 17:15 |
|
Lmao we defeated ur AI by training it to detect normal people and then, at the critical moment, acted like complete retards.
|
# ? Jan 18, 2023 17:16 |
|
*casting the mimic spell in ds*
|
# ? Jan 18, 2023 17:16 |
|
KirbyKhan posted:Lmao we defeated ur AI by training it to detect normal people and then, at the critical moment, acted like complete retards. surprised that the marines were able to keep themselves normal for six days
|
# ? Jan 18, 2023 17:19 |
|
KirbyKhan posted:Snake! Snaaaaake! SNAAAAAKE!!!
|
# ? Jan 18, 2023 17:19 |
|
bvj191jgl7bBsqF5m posted:computer_genious.eth: I believe that AI is an autonomous being that can one day learn to feel emotions jesus christ tho. way to figure out what is scarring to human brains by literally scarring hundreds of human brains, but in a test setting. unit 731 poo poo. ill be passing this on to a couple of people
|
# ? Jan 18, 2023 17:21 |
|
lollontee posted:jesus christ tho. way to figure out what is scarring to human brains by literally scarring hundreds of human brains, but in a test setting. unit 731 poo poo. ill be passing this on to a couple of people It's an extension of what large scale social media moderation ends up being. There's a lot of "content control farms" or whatever fancy words are used for them out there, and they're all as horrific as this, with pretty much the same job description, horrible pay, and resulting mental issues.
|
# ? Jan 18, 2023 17:40 |
|
Semi-related to the above: Beyond the obvious issues of techbro bullshit, the AI hype runs into the same issues that previous automation had. What is touted as "substitute for human labor" ends up being a mere multiplier on labor. Although it tends to transform the form which that labor takes.
|
# ? Jan 18, 2023 17:48 |
|
So Darpa chief got heart attack gunned in MGS1 because he failed so hard at making the genome soldiers recognize cardboard boxes as potential enemies
|
# ? Jan 18, 2023 17:50 |
|
vyelkin posted:https://twitter.com/ZaneGTCooper/status/1615586981177708545
|
# ? Jan 18, 2023 18:06 |
|
can someone please ask AI hitler if he is in heaven? we can finally get to the bottom of how much forgiveness christ has in his heart
|
# ? Jan 18, 2023 18:08 |
|
you could also ask the ai to do programming for you
|
# ? Jan 18, 2023 18:13 |
why would you make this like, what was the intended purpose
|
|
# ? Jan 18, 2023 18:14 |
|
Dustcat posted:why would you make this raising your skill level in internet arguments
|
# ? Jan 18, 2023 18:16 |
|
KirbyKhan posted:Lmao we defeated ur AI by training it to detect normal people and then, at the critical moment, acted like complete retards. this is very funny to me because it basically works exactly like the inspection machines I work on at work. It doesn't know what a "bad product" looks like, all you can do is train it to recognize what the ideal is supposed to be, and then have it detect any variations from this ideal. and I guess what I'm saying is if the military is depending on this technology they better hire some manual inspectors to double check their work lmao
|
# ? Jan 18, 2023 18:17 |
|
Dustcat posted:why would you make this When you add a strong negative incentive to producing "Yes, I did X and I'm glad I did it, and I'll gladly say A, B, and C on top of that", while preserving the entire structure of incentives that produced it in the first place, you get "It is very regrettable that X happened, due to factors A, B, and C which made it seem like a good idea at the time". There was no 'intended' purpose, but a purpose was built in none-the-less.
|
# ? Jan 18, 2023 18:19 |
|
lollontee posted:raising your skill level in internet arguments tired: playing yahoo chess as the chess master wired: leading a fascist uprising as hitlerbot
|
# ? Jan 18, 2023 18:19 |
|
Dustcat posted:why would you make this I think the reason why people make funny photoshops is so that others can laugh at them
|
# ? Jan 18, 2023 18:22 |
|
|
# ? Jan 18, 2023 19:09 |
|
Pepe Silvia Browne posted:can someone please ask AI hitler if he is in heaven? we can finally get to the bottom of how much forgiveness christ has in his heart There is no crime that justifies eternal damnation IMHO
|
# ? Jan 18, 2023 19:21 |
|
Butlerian Jihad when?
|
# ? Jan 18, 2023 19:26 |
|
Terror Sweat posted:There is no crime that justifies eternal damnation IMHO hmm... i'd really like to know what jesus himself thinks about this tho...
|
# ? Jan 18, 2023 19:28 |
|
losing my mind at the hitler paywall
|
# ? Jan 18, 2023 19:32 |
|
I need to know if this is photoshopped (please don't lie to me, I'm very delicate like an orchid in a gentle breeze)
|
# ? Jan 18, 2023 19:33 |
|
|
# ? Jan 18, 2023 19:44 |
|
https://futurism.com/neoscope/ai-analyzed-personalities-secret lol
|
# ? Jan 18, 2023 19:55 |
|
bvj191jgl7bBsqF5m posted:I need to know if this is photoshopped (please don't lie to me, I'm very delicate like an orchid in a gentle breeze) It's hard to explain, but the digital reconstructions are mostly generic conversation scripts trained from generic chat logs. They sprinkle in some text from the profile and it changes tone and nouns, but doesn't really change sentence structure or flows of conversation. They arnt photoshopped, but they arnt disembodied posting ghosts. I got to help train and test a medium follower twitter person's gpt bot and she only had to feed the trainer like 4 or 5 tweet threads to get it to sound enough like her. Biggest revelation and annoyance was that whenever it was asked a political question it responded like a fuckin lib. Mealy mouth platitudes that mean nothing are the baseline for these chatbots and that makes me lol.
|
# ? Jan 18, 2023 20:02 |
|
Woke Mind Virus posted:losing my mind at the hitler paywall thread title
|
# ? Jan 18, 2023 20:07 |
|
Yeeeep. poo poo like this is literally just tarot under a different guise.
|
# ? Jan 18, 2023 20:17 |
|
|
# ? May 6, 2024 05:57 |
|
Terror Sweat posted:There is no crime that justifies eternal damnation IMHO yuh gently caress
|
# ? Jan 18, 2023 20:44 |