@Grey_Athena The anti-data center movement is rapidly building steam in a very grassroots fashion. If you want anti-AI hopium don't just read about it, get involved, I can virtually guarantee there's something in your area
English
CentipedeTolerator76
1K posts
Mass surveillance, climate change, AI slop everywhere worshiped by idolbuilding cultists, rising fascism, ecological collapse, the endless death cult of capital. hard to feel optimistic about the future
English
@UniteUSUK @nise_yoshimi The UK has basically been a vassal state for the US since WWII and if they did formally merge Britain would never be able to escape a subordinate relationship to CONUS. Haven't you already won?
English
it's such a fascinating mess, our identity, and it's all because The Empire neither began with a proclamation or ended with one. A vast quagmire of Britishness choking everyone in the isles who tries to understand themselves
yoshimi red@nise_yoshimi
the dilemma of the English right: "Britain" here refers to an ideal national white ethnostate which Britain, tied intimately to empire, has never been. They mean "English", not "British", but "England" is nothing - this flimsy UK/Britain distinction is their flimsy compromise
English
@DirtyHamasnik @CantEverDie A single person might be able to make an *internally consistent* definition of AGI. The issue is that AI critics, hypemen, researchers, companies all have wildly differing definitions *amongst themselves*, so any such definition will more than likely end up being personal
English
@DirtyHamasnik @CantEverDie Okay great. What's your definitions of intelligence and AGI then
English
my biggest pet peeve around LLMs is when people (usually those invested in its success) call it “intelligent”. it definitionally, how it functions on a base level, is not intelligent. the way LLMs are built, it can never hit real intelligence. it’s just predictive
English
@DaveCash0527 @M_Cottone @CantEverDie It responds to user feedback, yes. But it's *trained* on a gigantic corpus of text that, at this point, basically consists of the entire internet - not just user interactions. And there has been huge user pushback to overkill sycophancy (see the whole GPT 4o shitstorm)
English
@CentipedeAccept @M_Cottone @CantEverDie It’s a fundamental flaw because it learns based on user feedback and users generally react more favorably to being told they’re right than being told they’re wrong. I’m not saying they’re completely useless but if they’re being relied on for knowledge gathering that’s dangerous
English
@CantEverDie I mean there's also the issue nobody knows what the fuck a good definition for "AGI" looks like. Similarly to nobody being able to agree on "intelligence" or "consciousness". OpenAI's definition of AGI is literally "can do X percent of economically valuable tasks", lmao
English
whenever you see people be like “its close to hitting AGI” just know they are lying to you because it’s genuinely currently impossible for this to happen at all
English
@DaveCash0527 @M_Cottone @CantEverDie This is more a property of how LLMs are trained rather than a fundamental property of every LLM. You can tune a model to disagree with everything you say, too. FWIW more recently models have been improving at "bullshit detecting", it is definitely still an issue though.
English
@M_Cottone @CantEverDie The problem is that it can’t produce outputs that match those an intelligent entity can, because it’s functionally designed to agree with its user even if they’re wrong. It will literally give you incorrect information if you prompt it to, knowingly or otherwise
English
@PovialJunk @M_Cottone @CantEverDie A deaf from birth person can't magically extrapolate what sound is like, for example. Humans aren't some magical outlier from "process input, produce output", and I guess we can debate but I don't think we have some sort of divine intuition. Just instinct and recombining symbols
English
@PovialJunk @M_Cottone @CantEverDie "fundamentally cannot extrapolate outside of the data it already has" can humans do this? Don't get me wrong we are extremely, extremely good at recombining things we've already perceived into new forms. But you still can't imagine a color that doesn't exist.
English
@PovialJunk @M_Cottone @CantEverDie That's due to the weird quirk of the tokenizer system (LLMs process text in batches, usually word/syllable, not letters) not because of some fuzzy "they can't read they can only predict" thing. If LLMs read character by character they'd ace the test every time, but they don't
English
@M_Cottone @CantEverDie that’s why we’re however many bajillion dollars into AI and they can’t reliably answer “how many letter [X] appear in the phrase [Y]” because it can only predict how many would make sense, it can’t actually read
English
@CantEverDie Like, it may not be 'really intelligent', but if a company can lay off 50% of their workers those 50% of workers are still laid off. What do you expect to happen? Someone to jump in and say "no, you can't do that, because it doesn't fit my pet theory of intelligence"?
English
@CantEverDie Also LLMs don't have to hit some arbitrary, fuzzy, subjective goal of "real intelligence" to displace the labor market or have massive consequences on software, warfare, medicine, etc. At some point trying to lawyer our way around these incredibly slippery terms is a distraction
English
@CantEverDie The big issue Alan Turing ran into, which we're still litigating somehow, is that if you ask 500 different people for their definition of "real intelligence" you get 500 different answers
English
@chow_global @cszabla We admittedly do a shit job of conveying how important this stuff is, but you'd think "find the area under a curve" is pretty obviously important.
English
@cszabla he cannot connect integrals and derivatives to his daily life? are you kidding
English
one of the reasons math is the most humanities of STEM subjects is mathematicians have to put up with shit like this e.vnexpress.net/news/news/educ…
English
@JaycelAdkins How many pages would you read if you spent all the time you spend doomscrolling/situation monitoring/watching short form video/netflix slop with reading?
Willing to bet it's actually *more* than 400 a week. And obviously you would learn more
English
@JaycelAdkins This is like 40 pages a day is this really masochism level for people? Maybe if you have 30 minutes of free time a day and you're using it to speed-read. But no, that doesn't seem like a lot. Totally reasonable amount to read for pleasure
English
Do people read 200-300 pages per week for pleasure?
English
@supercollider_x @michaelmiraflor We're social animals and if you can't recognize that you might be too far gone, man. Even introverts need social interaction. Come on
English
@michaelmiraflor i think because a lot of mainstream normies got in
comp sci became prestige maybe? status seekers went in, they care less about the ideas and tech and just want money
normies actually like socializing
English
It’s like an entire tech generation has completely forgotten all of San Francisco history prior to the invention of social media. Shame on you, nerds.
pnorm@paleonormie
why is SF so anxious about developing a "party culture" anyway? Its an expensive city for former gifted kids who work 70 hours a week in tech, it's not designed to be a party town
English
@Austen Lmfao 'communist drum circle' look at all of the grassroots anti-data center stuff happening in Appalachia
English
“I think SF tech bros don’t realize that all of the people in my communist drum circle are representative of the average American, and they all hate AI and think we should firebomb data centers.”
English
CentipedeTolerator76 รีทวีตแล้ว
Society really should have better answers than just "kill yourself" or "infinite rent-seeking luddism" for people who lose their jobs due to big market changes
More Perfect Union@MorePerfectUS
NEW: If Waymo gets its way, 2 million workers will be out of work. When Waymo gets a firm hold on a city, wages go down. Some drivers now have to work 12 hours day, 7 days a week just to get by. This isn't inevitable — but Big Tech is spending millions to make you think it is.
English
@PatBryanTX2 @ArchysLife Is Trump really bluegrass fascism? Trump is more bro country fascism to his fans, and Andrew Lloyd Webber fascism privately. Bluegrass was definitely 'hippie/woke' music at the time, its creators were very open about their inspiration from Jazz/Blues and held progressive stances
English
@ArchysLife However, Trump's Bluegrass Fascism and Orbán's Polka Fascism and Hitler's Wagnerian Fascism were all played in the same key.
English
I keep touching this point. Trump and MAGA are more classically American than any post-WW2 president or movement - the con-man, the quack doctor, and the circus entertainer are all quintessential figures of American folk history.
Derek Thompson@DKThomp
It’s interesting how many aspects of modern political commentary hold up the 1945-1971 postwar period as the natural state of things that was broken by our weird new modernity, when instead maybe it’s more accurate to see this period as profoundly unusual. I think about this with media commentary all the time: “Why can’t we get back to Walter Cronkite, shared sense of reality, etc” A brief and strange information oligopoly created a scarce number of radio/TV stations, which enforced a news monoculture on radio/TV audiences. Whether that was altogether good or bad, it was extremely weird! Look at the 19th century. A zillion newspapers, many of them insane and terrible and partisan. The chaos is what’s normal.
English