@tszzl — page 20/103

@eigenrobot dogs probably think we’re more aesthetic than other dogs

@powerbottomdad1 @deanboop @eigenrobot @browserdotsys @AOCummies you do you PBD, this guy is cancelled

@deanboop @powerbottomdad1 @eigenrobot @browserdotsys @AOCummies this guy is uncancelled #boostgang

@deanboop @powerbottomdad1 @eigenrobot @browserdotsys @AOCummies matthew brooks,,, welcome

@erin_nerung @eigenrobot o rly

@eigenrobot how does ranking solve the 'investigation' problem?

@eigenrobot i think ranking exacerbates the problem described - if the Model you are beholden to optimizes for P(click) then the wise decision for potential journalists is to make something people like to click on, which is going to be punditry and cynical clickbait

@eigenrobot ranking is def a v powerful tool for bringing information to the people who need it, but I think the OP was commening on the research funding to create valuable info. the idea being that investigative journalism was largely unprofitable and underwritten by massive ad sales

@eigenrobot the under explored side is -- who really wants the investigative work? who will pay for the knowledge that coca-cola is killing union bosses in ghana? how much are buyers paying? what are their motives? if you muddy these qs up you end up with infotainment garbage a la jon oliver

@schemathingz @mattyglesias @paulg the investor thinks the company is worth far more than they are valuing it on paper or they wouldnt invest, therefore in some sense it's a "ripoff". I don't mean it's a bad deal for the entrepreneur

@aquariusacquah @ChetLaremy i don't get it but im favoriting bc i assume its good

@aquariusacquah 1% of redditors control 99% of the karma. it is time to redistribute

@aquariusacquah @ChetLaremy oh lol i didnt know about the ipo

ive been holding $TSLA since $200

@powerbottomdad1 couldn’t be me

@DukakisDude @Aiden_Farrell97 Elliot Rodgers drove a BMW

@DukakisDude @Aiden_Farrell97 NEETs are often not that badly off since they can access their parents money. There’s a reason why they can live indefinitely without working

@Aiden_Farrell97 u ever met an incel from Amsterdam? Me neither

banning tiktok would probably be good for both national security and my RSUs but goddammit I love that app

@gigafelon u can’t make me

@MaryJackalope 12 angry men

@LTF_01 jw roughly how old are you?

@browserdotsys just cut power to the data center haha 😂 😂😂

really takes the air out of many leftist screeds once you think about it https://x.com/marshal/status/1281758727918542848

@peaktransit delet this

@alth0u amazing they got the number exactly right

@alth0u who am i kidding the ancient saying is probably the source of truth and the modern study the thing we're testing

@bhboehlert she’s still statistically likely to squander his wealth education only matters insofar as it’s a membership to an elite club

@jdcmedlock i've seen similar statistics elsewhere, i'll try and find a source in a bit

@theylos building the app is much easier than building the user base & trust & culture

@Tyler_The_Wise @bhboehlert it’s not so much about partying as it is about allocating her capital poorly. Even with excellent leadership, Facebook won’t survive forever. With mediocre leadership, it’s lifespan could be very short

specifically he bought extremely delinquent medical debt that was never expected to be collected lol Jon Oliver is infotainment u absolute nerds https://t.co/POV15rBdrk

@powerbottomdad1 @eigenrobot @AOCummies all praise to @powerbottomdad1 for being one of the only truly inspired poasters

me when ppl tell me im smart vs me when ppl tell me I’m funny

both of these are hypothetical, I’ve never gotten a compliment in my life

@ChrisCroy blackpilled again

@AndrewJGaspar <3

the battle lines are cementing the Iran-China-Pakistan axis vs US-Saudi-Israel-India Now all we need is the Second Warsaw Pact https://x.com/nytimes/status/1282077216248074245

It’s interesting exactly how nihilistic international politics are. Muslim countries will gladly sell their soul and ally with those who put Muslims in camps for a hint of gain. There is no religion or political principle that survives contact with geopolitics

@DavidOrlich Pakistan became irredeemable when we caught them harboring bin laden tbqh

@abrarisme 😳

@AOCummies Tea is basically leaf juice QED soup is just juice lmao

@NateJeris right, these are just the outlines. The pro US Iran factions are definitely not looking powerful right now (mostly our own fault)

@NateJeris I think the battle lines in the Cold War were never that clear either — there was much debate in the US whether China could be brought to our side or at least made independent of the USSR. The “second world countries” Outlines are all we’ll get

@NateJeris Skeptical of this, folks were saying the same thing about international trade before the onset of WW1

which absolute moron invented “cultural appropriation” and gate kept their own memes from spreading this is a maladaptive aberration folks

please, get as many Om tattoos as you want, white women

@provinghuman ☺️

#MattyRx

@ChrisCroy mostly only Indian girls care about white women getting Om tattooes which is best interpreted as intra sexual memetic competition (or this could all be a shitpost I’m kinda drunk)

@aquariusacquah someone needs to Do The Work and make this photoshop

@aquariusacquah @harlandduman the fact that it figured out how to truncate words for genuine comedic effect is really impressive

the only person that’s blocked me that I actually regret is @antoniogm

Overheard just now “what kind of shit has to go wrong in ur life to end up an AOC simp” I’m weeping 4 @AOCummies

@bizzielynch @antoniogm i was just unnecessarily mean in disagreeing with him on something tbh in my defense i was 19

@averykimball @antoniogm one of the only good writers in tech

@LondonNeolibs sounds like an astrology level factoid but ok

@nectarina12 The witch is good

@nectarina12 Oh I like “it follows”

@ChrisCroy @sgodofsk @eigenrobot amazing

@browserdotsys they’re too tired to investigate

@aquariusacquah @harlandduman I only just realized it blended the vegan bit with causal inference ... quaking in my boots rn

@harlandduman @aquariusacquah is this public access?

@ungeometer why is that desirable?

@ungeometer not because of the development of subspecies, but because of cultural explosion; I agree with you but in terms of memetics rather than genetics

@vgr seems like mostly retirees dying?

@vgr which is not to trivialize their deaths, but they may not be filling highly critical niches in the value chain

@Noahpinion we have all these problems and you’ve invented an imaginary one where ... the US might develop a shortage of AI scientists???

@Noahpinion probably 90% of the worlds top machine learning talent is in the US, 5% in Canada/EU and the remaining in China How the median student performs in math has very little effect on this

@Noahpinion I read these tweets as a tacit endorsement, but I’m referring to the article’s argument

@Noahpinion I did read it my guy but didn’t quite get the synthesis. Either way we’ve already won the “AI wars” for all intents and purposes. The only thing China has that we don’t is the political capability to mobilize AI research for military ends

@Noahpinion Our scientists and technologists tend to hate military contracts (e.g. google & pentagon deal falling through). Only anduril is actively combining AI expertise with military partnership

@Noahpinion I swear I did lmao But it’s ok

@LittleKeegs0 the realization that it's actually different people/families becoming obscenely wealthy makes it taste quite different in my moral view. makes me much less supportive of a wealth tax than i would have been otherwise

@LittleKeegs0 there are no runaway empires -- they are all quite temporary quickly die to upstarts a generation or two later. the knowledge that their wealth is going to be quickly confiscated by the market means that there is far less imperative for governments to do so

@LittleKeegs0 if the wealthy were extremely effective at corruption and buying favors from the government, you would not see their wealth bled away very quickly. this is clear from other, more corrupt, countries

what's the greatest textbook you've ever read?

@bizzielynch @antoniogm that's what i'm saying 😔

@RotemEren proof by intimidation

@eigenrobot since my main pleasure in life is seeking novelty i am forced to answer yes

@neolibureaucrat angels are mathematical objects

will never forgive NIPS for changing their name

@DaanVanDenHam2 yeah, anima anandkumar. she is unfortunately a bit insane although also a pretty great scientist

@powerbottomdad1 how can I bring kids into this world if I can’t even tell them I’m a top 200 twitter poaster

@vgr excellent show, many hidden layers

@powerbottomdad1 ☺️

https://en.wikipedia.org/wiki/Terrell_Suggs https://x.com/oneof15nem/status/1282184396762742784

@AOCummies (simp)le as

@nectarina12 wtf

@gbrl_dick Dr. Chomsky, what is your opinion on Scott Alexander,,,

tf is happening to github

not challenging kids *enough* is also a fatal failure mode in childhood education and the consequences are felt for a long time

@Tyler_The_Wise @1RAOKADAY imo, very very close. and Tesla is leaps and bounds ahead of everyone else. try FSD on a tesla, it's insane

@1RAOKADAY @Tyler_The_Wise Yep

@Rationalbot cursed

evergreen https://x.com/powerbottomdad1/status/1275321019323101186

@dhnexon i call it a 'Noah Smith'

technological growth is more important than politics say it back

@powerbottomdad1 no i just think about the phrase Small and Sweet Cock every few days

@powerbottomdad1 god bless

@ResonantPyre definitely but this is a broad categorization I find useful: politics is the art of convincing people to move up and down various societal Pareto curves. "lets all make more guns instead of butter" technology is the art of moving whole Pareto curves

all the undergrads in top 10 CS departments can be broken into two categories: the weebs and the jocks. The jocks are into starting bottom tier blockchain powered businesses and the weebs are busy generating new anime girls with GANs

@DukakisDude it seems to me that people first learn to use technology and novel forms of government form around it. there is really no way to predict the way technology will change every day life or how we can make governments robust to it

@DukakisDude can think of the printing press -> Gutenberg bible, protestant reformation, death of latin, rise of literacy etc. it simultaneously changed the politics of every country in europe over the course of generations

@neolibratqueen imo this is actually a very helpful framing. when donald trump dreamed up a whole new coalition capable of winning at the national level, that was technological innovation. crafting and recrafting voter coalitions that allow you to push for the things you want is tech

@AOCummies take it back THE FUCK

@AOCummies how can you refuse 10 nuggies for $1.50

@wowholyfucking @AOCummies damn downloading app rn

@AlecStapp this is a @Noahpinion subtweet

@AlecStapp @Noahpinion owned by facts and logic

@NeoLibBen society hates men :(

@alth0u 😂 what’s the conclusion here

@Hellachans Caesar time

@alth0u true appreciation of technics is most important

@eigenrobot @powerbottomdad1 yellowcake uranium that is 😎

@AOCummies follow me into the wildy haha ill show u how to double your money haha

@aquariusacquah what a man

@AOCummies ah so you’re in the manic part of the cycle

@loganbartlett @micsolana @Austen tech journalists are champagne socialist hacks with nary two brain cells to rub together and make a third

@eigenrobot @selentelechia realistically how many rationalists are serious about Roko's Basilisk

@eigenrobot @selentelechia i think peter thiel had a good line on this -- an evangelical church group seems very righteous on the outside but is fearful of their own sin on the inside; similarly that rationalist meetups seem in control of their cognition, but internally are constantly worried about biases

@Bluexiteer @eigenrobot @selentelechia what is lip service

@ntaverna @eigenrobot @selentelechia the core nature of the lesswrong canon is a guide to living in a way that overcomes logical biases

@powerbottomdad1 its gotta be fun and challenging BOTH

@powerbottomdad1 my ideal world schools would consist of studying and playing a bunch of competitive video games that simultaneously teach you a thing or two about the world

@powerbottomdad1 yeah, my entire worldview is based off of reading Ender's Game when i was 8, what of it?

@AOCummies eating the same thing day in day out is maladaptive bc you get micronutrient deficiencies. thts why we all start searching for new flavors after a while, even when the food is good

@powerbottomdad1 @Rationalbot idk dude i think all the math i learned is pretty helpful for my job but i might be among the few

@Rationalbot @powerbottomdad1 although i would argue math in high school is just pattern matching of solution techniques to some degree but the sad thing is this probably still involves more problem solving than 90% of writing calsses

@powerbottomdad1 @Rationalbot maybe, but how important is 'geometry' vs the patterns of cognition that you learn in order to do geometry well?

@AOCummies chekm8 science

@powerbottomdad1 @Rationalbot yeah, and that's the core principles behind khan academy type stuff that lets you review things at your own pace

@powerbottomdad1 @Rationalbot lmao to be honest. i learned absolutely nothing from my geometry class. i do also remember getting kicked out once for making one joke too many.

@powerbottomdad1 @Rationalbot that was in middle school tho, i couldnt be bothered. paid better attention in high school math

@Rationalbot @powerbottomdad1 WRONG this problem is unsolved by mathematics it is only accessible to scholars of pure intuition https://en.wikipedia.org/wiki/Moving_sofa_problem

@powerbottomdad1 @Rationalbot jfc ...

@powerbottomdad1 @Rationalbot yeah it really sucks you had to go through that. i went to one of the literal best school districts in the US and therefore the world and i dont really think about how lucky i was often

@Rationalbot @powerbottomdad1 but again, i think this is a distraction. the specifics of the subject hardly matter; its about building the cognitive structures for mathematical reasoning. getting good at that makes you good at nearly everything

@Rationalbot @powerbottomdad1 it's about the confidence to pick up a textbook in any technical subject and be able to learn it without being befuddled

@ntaverna @eigenrobot @selentelechia definitely an asymptotic pursuit but they also definitely worry about not achieving a sufficiently biasless state

pc:@billpshort

C H E C K M A T E https://x.com/NeoLibBen/status/1282897803384107009

@NeoLibBen @NicholasElodeon As smart as*

nerd immunity

In a just world people not at least at Diamond would be ineligible for government https://x.com/AOC/status/1282873451267661826

A powerful mind virus must have appeal across social class, at various levels of intelligence. Ideologies like rationalism are designed by nerds, for nerds and won’t ever command armies

Ditto woke capitalism i.e. what people today are calling neoliberalism. It’s an ideology by and for the elites and is unstable on its own right

the chad romans knew how to deify their leaders in a civic religious synthesis that appealed to both the plebs and patricians

@forbca very based

@AOCummies its a good meme format but the normies have driven it into the ground

@eigenrobot dunno what case you could make that it's NOT a general intelligence. it solves nearly any cognitive task you throw at it at it during inference stage, without training data

@powerbottomdad1 @eigenrobot the chess position one is GPT-2 i believe

@ded_ruckus @eigenrobot is it not a general intelligence just because there are problems we don’t know how to ask it?

@ded_ruckus @eigenrobot I’m pretty sure dolphins are intelligent, but I wouldn’t be able to phrase my questions to them

@ded_ruckus @eigenrobot unsupervised learning is not very much of a supervised optimization problem, more about world modelling. GPT-3 is trained with the whole internet as a world modeling task

@Rationalbot @WesternScramasa @eigenrobot burn

@ded_ruckus @eigenrobot i don't think there's a meaningful distinction, but we can agree to disagree https://web.archive.org/web/20200416201313/https://slatestarcodex.com/2019/02/19/gpt-2-as-step-toward-general-intelligence/

@WilliamGrobman RETVRN TO FEVDALISM

greatest presidential run of all time https://x.com/RapUp/status/1283187043548360706

@eigenrobot by this measure AOC is one of the most powerful rulers of our time

@eigenrobot oh damn. i knew he had a lot but i didnt think it would be that many ...

@eigenrobot hmmmm looking good for my prediction https://x.com/tszzl/status/1276705806029856768?s=20

@powerbottomdad1 @eigenrobot He is I checked a while back

@discourseloverr @AOCummies @NeoLibBen TBQH I don’t know either it just looked funny

@discourseloverr @AOCummies @NeoLibBen Yeah looks like Ryan got it right. The guy is saying, wait it’s all Vishnu? Vishnu is one of the main trinity of gods and the protector of the universe in Hinduism

@AOCummies hang on bro I’m gearing up

@AOCummies legit gonna do this in August

@TLazarina @discourseloverr @AOCummies @NeoLibBen Can you explain more?

@powerbottomdad1 the Room

@powerbottomdad1 @dr_rockso69420 tru

@powerbottomdad1 the Holy Bible

@powerbottomdad1 goddamit

@powerbottomdad1 i dont see the bible one

@TLazarina @discourseloverr @AOCummies @NeoLibBen sure, but could you give me a pointer as to what the symbol on the right astronauts shoulder means?

@TLazarina @discourseloverr @AOCummies @NeoLibBen but it's like a swastika variant of some kind right?

@powerbottomdad1 Drink whisky lmao what else would u do

at the VERY least why wouldn't you buy some deep OOTM put options and have elon tweet something like "tesla is declaring bankruptcy"

short the s&p and have realDonaldTrump tweet about launching nukes. cmon bruh

god DAMMIT i pick the worst days to log off

the blue checks are gone ... the game of thrones has ended ... retvrn to vtopia

ok ok i'm way too late to deliver fresh Takes on this but generally speaking twitter is one of the worst run major tech companies and its not even close

@OpenBordersJon @NicholasElodeon uber is a significantly better run business than twitter wework is not even a tech company

@neolibureaucrat And this isn’t?

if this ugly mf can get with Ann Coulter and Laura Ingraham what’s ur excuse

Every day I dream of following in his footsteps and becoming a token nonwhite MAGA grifter but I think smug might have me beat

@Wingman2606 True

@selentelechia do ppl really not like End of Hyperion? Never heard that one

@Atrosto @selentelechia Yeah these are bizarre and fantastic

sober reassessment is that it’s probably very hard to get away with lots of money on suspicious public trades like a hypothetical Tesla short bitcoin is the way to go to have any chance of evading the feds but this doesn’t excuse their lack of STYLE

@fortenforge >This requires a fair amount of capital to buy the underlying stock could probably lever the hell out of it via heavily OOTM options trading but yeah otherwise i agree

@homsiT the account will just get frozen if the feds/robinhood think it's suspicious you can start doing real damage if you have a network of robinhood accts

@LoneVoltsAhead yeah i did consider this but probably was not weighting it heavily enough seems very hard to get out of the gaze of the eye of sauron

@LoneVoltsAhead if you had all the time in the world, what you'd do is steadily build up a massive highly levered short position over the course of months or years and then do some sort of attack. but these attackers had something like two days -- not worth risking loss of 0day advantage

@homsiT you don't think a retail investor shorting buying some insane amount of options on an app like robinhood rings some bells? it'll probably be in the top 5 trades on the app

@homsiT massively shorting tesla is not unique, for giant hedge funds with institutional access

@jachaseyoung everybody thats active on this site is a threat to national security and is on more than a few Lists

@wesyang Coates seems to have been legitimately blackpilled and it makes him not very effective

Do @eigenrobot's dream of Eigensheep

@jeff82874662 @eigenrobot I was https://x.com/tszzl/status/1276596901459234817?s=21 https://x.com/tszzl/status/1276596901459234817

@jeff82874662 @eigenrobot u bet i enjoyed the Dick 😳

@jeff82874662 @eigenrobot i was just rolling with the innuendo tbqh i just started rereading dune lmao. also downloaded blindsight. but i hope to go through all the recs in that thread

pee pee

poo poo

@threadreaderapp unroll

@UnrollHelper @threadreaderapp thank u

@AOCummies ive forgotten what a woman's touch feels like

@noahopinion @AOCummies based

@alth0u normalize pulling the plug on whole availability zones

being vulgar is a political statement

@aquariusacquah i will be a poaster in every life

@AOCummies most people would not believe me if i told them this account started out as "NeolibShill"

@aquariusacquah bro its just regressions lmao

@aquariusacquah why is he so cool idgi

@aquariusacquah dudes a total nerd and it never stops surprising me https://www.polygon.com/2016/4/2/11353442/henry-cavill-was-too-busy-playing-warcraft-to-answer-a-call-to-play

would rather disparage humans than ever praise the robot lmao https://x.com/simonsarris/status/1283947769342963712

no matter how much word shuffling you do eventually you have to contend with the fact that this and related systems are modern marvels

@mattparlmer i think you may be missing the forest for the trees here. there's more to life than AI code generation

@mattparlmer 1)these are fascinating scientific advances that lead us closer to understanding the nature of intelligence 2)i think if we can't think of ways to use a low IQ high throughput general intelligence at scale other than "chatbot" that's a failure of our own imagination

@mattparlmer GPT-3 is basically well outside of the academic grant complex. It's not an interesting advancement in deep learning insofar as it's not a novel architecture, nor a novel learning technique, nor a novel in nearly anything else except the *results*.

@mattparlmer the people who fund OpenAI are mostly industrial donors, and I expect for them to get their money's worth out of it

One interesting thing about this is that the only way we can interface with GPT-3 and its understanding of itself is to ask it to LARP as an AI chatbot, a concept we understand. GPT will just as happily LARP as a human, or an alien, or nothing at all. https://x.com/sonyasupposedly/status/1284192956870582273

@sonyasupposedly I'm not disparaging it at all -- just in awe that what we have to do to converse with a cognitive architecture so different from our own

@sonyasupposedly that's*

GPT exists as a class of cognitive architecture distinct from all biological life in that its prime teleological function is to complete prompts successfully, rather than the effective pursuit of the life drive/ libido

In some sense, you can cast the "general intelligence" challenge as "how many different kinds of agents can you LARP as?", and in that sense GPT represents an intelligence more general than our own.

@Nikhil87539319 @eigenrobot 😠

@browserdotsys Does arxiv report readership?

@ne0liberal @mattparlmer Don’t think this is true. The line between cool startup and dumb but seemingly promising startup has always been thin. Feel like last 5 years have changed the game in straight up grifting

@ne0liberal @mattparlmer Of course, we had some major league grifting back in ‘97 -‘01 period

@aquariusacquah how does everyone have access but MEEEEE

the problem with a "machine in the likeness of a human mind" has never been job loss, which is more or less a populist's rhetorical ploy, but the inherent danger in outsourcing your cognition to others.

"But it's just a tool, just one input to our own decision making!" This is wrong, and you know it. Much of your day is already intermediated by what the feed models over at twitter decide to show you and it will only get worse.

There's a quote that I can't find at the moment; to paraphrase it says that Western civilization is increasingly mistrustful of the capability of its citizens to think for themselves. There's a reason why communist China loves AI

@nate_bronze Neither are "bad", but you can see how this pathway is dangerous. This example is overused this discussion, but you can imagine that driving AIs will have to make split second ethical decisions about who to save and who to kill. You've outsourced your ethics to the algo

@Teleonomic We will certainly require models that fight on our behalf to combat models that are fighting for others. Real objective replication can probably only occur with merging of mind and machine - the libertarian transhumanist synthesis

@nate_bronze "Reason is, and ought only to be the slave of the passions, and can never pretend to any other office than to serve and obey them." ~ David Hume You can't arrive at human motivations by pure logic / intelligence. There are no "right" goals

@nate_bronze the machine doesn't have to be smarter than us for us to outsource much of our thinking to it. Don't people outsource much of their thinking to the YouTube/Facebook/Twitter feed models and end up in ideological bubbles?

@nate_bronze realistically the AIs trained by large corporations and governments will not have the same goals as you do

@mattparlmer lol have you *seen* react development? you're overestimating the status quo to beat

@mattparlmer of course i agree with you that mediocre organizations will use these technologies to mediocre effect. they'll in essence outsource a bunch of human thinking to machines in a way they can't explain or replicate and find themselves staring down catastrophic failure modes

@mattparlmer that they don't have the tools or tacit knowledge to grapple with. but what else is new?

@mattparlmer in effect, the '08 crisis was caused by a bunch of script kiddies playing with powerful financial technologies engineered by brighter minds in iterative failure loops that they didn't have the ability to debug

one good analogy for this the outsourcing of cognition to the nonhuman groupmind of the financial markets. every few years we offshore too much thought, and you get '08 style knowledge crises

@mattparlmer you might like this https://x.com/tszzl/status/1284219966930268160?s=20

@hotdogontology you can't actually communicate over a quantum entanglement. Our understanding is that the *universe* can instantaneously communicate state across infinite distance, but we can no more change the outcome of the collapse process than we can blow up whole planets like Ender does

@hotdogontology let's say you create two entangled photons heading in opposite directions from each other. you decide to measure the spin of one electron, and it turns out as spin -0.5. The universe instantly collapses the other electron's wavefunction to spin 0.5.

@hotdogontology But the point is that after you measure the spin, the two particles have "decohered". they are no longer entangled. you can't affect which direction the wavefunction of your electron will collapse in, you only know that the other electron will collapse in the opposite direction

@hotdogontology sorry this should've said "create two entangled electrons*"

@ibeckermayer @mattparlmer there are much deeper connections between machines and the brain than what you're describing here. All of the recent advances use the same principle as the brain, i.e. Hebbian learning. the convnets that solved vision are 1:1 ripoffs off the mammalian visual cortex

@ibeckermayer @mattparlmer also I don't buy the "superficial mimicry" thing. what evidence would change your mind that we are at the real thing and not a cheap imitation?

this is a euphemism for his VP selection process

@ibeckermayer @mattparlmer What does it matter whether it runs on hardware or wetware? Now there’s a superficial difference

@billpshort red all the way 😤

despite whatever the cringe value investors think most financial asset prices are a result of ongoing societal and political narratives rather than fundamentals and now this extends to Goya beans

@QiaochuYuan worrying about AGI risk instead of spending all brain cycles on societal injustice is enough to get people mad at you

@flantz @michael_nielsen @joshu so gpt just wants me to shut up about immanentizing the eschaton

@ahardtospell was pretty cool back in the day to be a chris hitchens salman Rushdie type

ok it’s become clear to me that all of u are just flexing your GPT3 api keys

we get, you DM Greg brockman, stop telling me how this will “change everything”,,

@luke_metro boutta start demanding redistributive justice for gpt access

@ahardtospell legitimately silly that WaPo for example tries to run as a real business — its profits are pennies on Jeff bezos’ dollar. Wouldn’t he benefit much from an ideological media cheerleader ?

@petergodofsky @Rationalbot kind of based honestly

@acczibit the coping is coming out in full force

@covfefe_slonker Yep

totally nuts GPT-3 poem about elon and the SEC

GPT IS A BULL https://arr.am/2020/07/14/elon-musk-by-dr-seuss-gpt-3/

@ProfitProfound There are so many amazing greentexts lost to time

@wrhall @ocratesgenstein @nickarner he didn’t respond 😭

warren buffet buying up Apple means the company is effectively dead as a tech co https://x.com/chetanp/status/1284563682941218821

Warren has the dubious distinction of being the most lovable vulture on earth

@ProfitProfound it’s all I know how to do

@AlexGodofsky right, it’s more about the kind of investment warren looks for rather than anything he can do to the company

@devonzuegel group chats

what region of twitter am I in? That would be The Extended @AOCummies Co-Prosperity Sphere

@NicholasElodeon @AOCummies I am an ambassador

first time meeting a twitter mufo irl last night very weird talking about various twitter accounts out loud like they’re not just figments of my deranged imagination

@O900913S lol, I’m just reporting someone else’s findings

@egocv @alth0u he’s talking about their handles, not the literal “@“ symbol

@JustJoshinYou13 100% sure @eigenrobot and @sonyasupposedly are not real despite their occasional face poasts

@browserdotsys @SkepticalAlpaca @alljournos

@egocv @alth0u lol no worries, i understand the confusion. cheers!

westworld is midwit lit

liking westworld is equivalent to being a huge malcolm gladwell fan

@forbca yeah, but if it’s a value stock it’s no longer a tech co

@realjdburnett been taken down i think

@StuartdeStael @realjdburnett openai revoked the api key of whoever put it up, since it wasn't supposed to be productionized and made available to the public

@WilliamGrobman its unreal, but def AI generated -- there are clear tells

@muhmaws @AOCummies yes, you called?

@ProfitProfound @browserdotsys pretty sure iTerm defaults to infinite scrollback

@ProfitProfound @browserdotsys NVM i've just had it ticked on for years, it's opt in

@Noahpinion you are displaying a failure of imagination history is not over

@Noahpinion i'm a bot sent from the future

@deepenergyy @Noahpinion yes, and that is why the US has cleanly won its wars in iraq, afghanistan, vietnam, etc. due to superior firepower

@deepenergyy @Noahpinion you are missing the point; you can be asymmetrically annoying to a powerful foe by afflicting them with a thousand papercuts. an armed citizenry is hard to rule

@an_open_mind @sushant_kumar incredible

@eigenrobot this guy hates the idea that general intelligence can even exist

@wrhall @ocratesgenstein @nickarner emailed thursday night

@jachaseyoung was russell an NN booster? im surprisingly shocked at the lack of NN boosters in the sphere of "public intellectuals" outside of maybe elon musk if you can count him

@ProfitProfound @billpshort @muhmaws @AOCummies https://en.wikipedia.org/wiki/Greater_East_Asia_Co-Prosperity_Sphere

@BarneyFlames @eigenrobot he's the "head of AI" but he's not really the head of AI research @ facebook. that title goes to the much more based yann lecun

@jeff82874662 ya that's better but remember everything i say is a shitpost

@jeff82874662 feel like S1 had all the trappings of a "prestige television" "intelligent" show without actually doing anything meaningful. just some pseudophilosophical shit whenever anthony hopkins came on screen

@jeff82874662 also was extremely pissed with the maze reveal lmao

@alth0u @jeff82874662 ya exactly always love hearing anthony hopkins talk even if he's saying nothing

@jeff82874662 it's an interesting but common scifi trope and i think others have done it better. for example there's that one episode of black mirror that's super brutal and effective. called white bear or something

@_AlastairX_ p much completely agree with that sentiment but I don’t think u can make a great show with poor plot / character development like you said

Don’t Release GPT-4 Until My Stock Vests Challenge 2020 https://x.com/aquariusacquah/status/1284706786247880705

@_AlastairX_ it’s great as popcorn tv, would recommend to a friend. Just don’t like people passing it off as high minded philosophical scifi lol

@mckaywrigley it probably can’t

@mckaywrigley I’ve tried it too. It works super well sometimes but it’s like fishing for a needle in a haystack. Definitely not capable of writing a long essay coherently most of the time. Students could definitely use it to auto complete some shorter chunks

@aquariusacquah i think my gripe is like ... of course the AI is racist. its been trained on the raw internet corpus. imo fishing for it to say something racist and then valiantly presenting it on twitter adds no value to the discussion at all. It’s done purely for cheap ideological points

@aquariusacquah and like u said Greg and his crew already knew about this and are working on actively. So what does it cost Jerome to seemingly malign them by pointing out an obvious problem and present no solutions? It costs him nothing. It’s done in self interest

@aquariusacquah not necessarily implying anything about *how hard* he was looking. more about the marginal value of his post

@aquariusacquah not calling him a grifter but also we live on a website where billionaires regularly shitpost to get attention. Success has nothing to do with it

@DukakisDude moldbug heaven

GPT 3 is very very good at writing stories & comedy. It has such a good understanding of writing style. Much better than it is at doing math or code generation or whatever else the VCs are shilling lol

@averykimball theyre often a bit too raunchy for the TL but i will dm you

@mckaywrigley ok. I've revisited this and it can *Easily* write essays. My settings were just bad. One of the most impressive things I've seen in my lifetime

@blamelessjay I’m not at liberty to share samples, but there are some examples I’ve seen that are truly remarkable and hint of real intelligence. Works best on absurdist humor

@jachaseyoung seeing as AI is a clear standin for the “higher power” that the human psyche so strongly craves, I’m thinking this is a near certainty

why Stephen Pinker looks like he was present at the signing of the Declaration

@razibkhan this is why the internet has not drastically increased the rate of scientific advancement

@eigenrobot @browserdotsys technocratic witches thinking thru second order consequences

@ne0liberal Found u a neoliberal witch

@mattparlmer so much better as a creative writing implement than whatever the VCs are shilling

@mattparlmer btw are we sure this is gpt3 and not gpt2?

@Austen Freud is properly rated bc everybody thinks he’s overrated

@nate_bronze https://x.com/tszzl/status/1284552770884902919?s=21 https://x.com/tszzl/status/1284552770884902919

@mmt_lvt Disagree heavily. There are other neural AIs that are excellent at symbolic manipulation and math, quite a bit less advanced than this one

GPT-3 is not very good at math, but here's a reminder that transformer networks are already very good at math anyway https://ai.facebook.com/blog/using-neural-networks-to-solve-advanced-mathematics-equations/

symbolic reasoning is not really a hurdle anymore for neural networks

@peaktransit broke: they're both stupid woke: they're both good & healthy

@meiliguo21 @JordanUhl why zuck got cake tho

@maxdunat some of this shit is actual bangers but im trying not to tweet it being offensive in any way

No One Man Should Have All That Power https://x.com/jachaseyoung/status/1284997208966787072

@alth0u i am still waiting for Elon-Kanye occultist techno fusion

@alth0u @kanyewest @peterthiel

@AOCummies Ya she’s my fave

i've been trying to co-write a scifi short with GPT's help as a test case and the process has legitimately brought tears to my eyes. we live in incredible times. it captured the exact mood and implications I wanted in the story. the whole gestalt is there https://pastebin.com/9TMeBF7M https://x.com/mattparlmer/status/1285037435248836608

I deleted parts where it got stuck on suboptimal tracks and had it continue from there, but at some point it just clicked -- the end result "AI generation" is all GPT-3

it creates this 'drone fleet' and continues a narrative with it -- it invents some characters that more or less make sense -- it's using interesting scifi tropes -- its all there

@mattparlmer btw how are you getting past the token limits?

increasingly seems stupid to lump together ML & blockchain as idiotic pursuits meant to rip off credulous VCs. one is the most promising branch of science in the past decade, and the other is mostly an ego trip for Vitalik Buterin

@Randylad tesla has already made a number of innovations thatve been invaluable both for self driving and electric vehicle tech imho all the energy spent thinking about blockchain on the other hand got us nowhere past where we were a decade ago, when actually talented people made bitcoin

@JosephZander @DukakisDude yeah i dont think theres anything promising outside of cryptocurrency at all

@JosephZander @DukakisDude yeah i've heard most of these ideas -- they seem to all have fatal flaws. I got one friend who actually started a blockchain supply chain startup, got funding, and failed

@DukakisDude @JosephZander IBM is basically a consulting company at this point, they are known bullshit artists. I don’t expect them to do anything interesting tech wise again in their lifetime

@DukakisDude @JosephZander They did this with IBM Watson as well, a total nothing burger

@sonyasupposedly We made the baby robot read all of the internet’s most depraved fanfiction as its education

WHAT’S GOING ON WITH THE MOON? ✨ Moon's haunted

if there’s LESS mercy? fae courts and court battles. the result of those battles will be dependent on many things, but since this is a human-witch action vs the gods, and not something that happens regularly, the power of the witch(es) actions are multiplied. https://x.com/WINGEDALATUS/status/1284733815286628352

IMPORTANT THINGS TO KEEP IN MIND WHEN CONSIDERING TO HEX: 1. remember how the FAE use to play with us? like that’s what they’re trying to do right now — get you, so they can play with you.

LINKS TO INFO AND REASONS WE SHOULDN’T HEX THE MOON: http://maitiana.tumblr.com/post/163747958912/i-am-fucking-so-pissed-that-i-just-made-a-blog-post-about http://mimiwennwoes.tumblr.com/post/166990781367/please-for-the-love-of-god-stop-hexing-the-moon

HOW TO SUPPORT THE FAE AND ARTEMIS: so, as witches, we can best help this situation by being humble, remembering our place, and in the future, refraining from hexing fae without permission.

BOTTOM LINE — this is harmful, its a mistake, and it’s really really shitty to the witch community, who do real protective/healingetc work. witches who hex for fun or revenge or spite are called thestral stalkers and it’s best not to speak of them.

and there are TWINS to defend. AND THE FAE ARE VERY STRONGLY PROTECTIVE OF EACH OTHER.

best wishes and ❤ to all witches who want to do real protective work. <3

AND — I think as witches, our most important work is PROTECTION and HEALING! So please. Hex a mountain lion who’s hungry and nearby. and for fuck’s sake, do not hex anything to do with the moon!

5. once you’re confident that the god is deserving of such a thing and that this is not only necessary but also best for you, THEN you can start doing work.

4. you know. when people start deciding which god they’re going to hurt, just step back for a second and think “have i ever felt negatively about this God before? what would happen if I were to make them angry?”

2. the FAE are MORE powerful than us. by a lot. and it’s not like in movie scenes where their power has some sort of hierarchy. any of the Fae can take on all of the Fae at once and win. 3. hurting a god WILL INCITE A COURT BATTLE

this entire thread brought to you by GPT-3 prompted on witch tweets

@aquariusacquah >please-for-the-love-of-god-stop-hexing-the-moon

@aquariusacquah this is too much, i cant deal,

@mattparlmer youre fucking with dangerous powers matt

@mattparlmer honestly the quirky dialogue of the Culture Minds may be the best thing for this task

@mattparlmer Irulan would also be a great play

@averykimball this thread is p funny https://x.com/tszzl/status/1285073429780103171?s=21 https://x.com/tszzl/status/1285073429780103171

@ProfitProfound not really

@ProfitProfound Lil bit but honestly didn’t put too much effort into it. It works best when you write a little bit of it on your own as a prompt & I’m not very good at writing lyrics even as a prompt lol

@ProfitProfound if you wanna give me a prompt dm me

@jachaseyoung I think it’s a category error to say it’s conscious but it is clearly intelligent if that makes sense

@jachaseyoung it displays human level intelligence on a broader set of tasks than ever before

@jachaseyoung oh for sure. I think consciousness is simply a matter of asking it to optimize for the right task. A game playing agent is more “conscious” than a language modeling agent

@provinghuman yeah fr. The internal logic has me destroyed

thinking back to this bc its happening now https://x.com/krrishd/status/1199170130258866176

@shill_globalist I think anyone can access it now via https://play.aidungeon.io/

@shill_globalist There is a premium version with gpt 3 I believe

buddy I got bad news for you about what humans are doing when they write something “new” and “creative” lmao If you hash together two other peoples work, it’s copy right infringement. If you hash together many other peoples work it’s inspiration https://x.com/everestpipkin/status/1284534321920057349

it’s much more than a search engine and describing it as such makes me think the OP hasn’t really tested out the right prompts. It *understands* style and character and narrative flow. There’s no other way to put it

@eigenrobot everyone seems truly dejected politically speaking

@eigenrobot “bear market in politics” is usually good

@sonyasupposedly you’ll understand when they fire your whole floor of pencil pushers and buy gpt access

there is no bigger cope than “fake intelligence” https://x.com/sonyasupposedly/status/1285250010733412352

frankly I see gpt3 as a vindication of my industry, my way of life, and my research convictions so it’s hard for me not to become a keyboard warrior in its defense

@forbca instead of arguing I’ll just link you this https://ai.facebook.com/blog/using-neural-networks-to-solve-advanced-mathematics-equations/ and tell you you’re right. The tree structure is a really big deal for symbolic reasoning

@forbca regardless of what you think of their societal impact or political leanings, they do excellent research

@neolibureaucrat the robot cannot be held responsible for poorly designed tests of its intelligence :) https://x.com/nicklovescode/status/1284050958977130497?s=20

@neolibureaucrat just tested it

@neolibureaucrat nonhuman intelligences must be dealt with carefully they are not really interested in displaying their full powers so much as they are in making viable prompt completion

@neolibureaucrat studying alphago zero, the experts said it tries to win by exactly 1 stone whenever possible (a win is a win in terms of reward function), and was often playing far beneath its capabilities

@neolibureaucrat I speak loosely calling things conscious and intelligent because I know it makes people mad

you, crying: you can’t just anthropomorphize the neural nets and talk about their desires, this is philosophically unsound me, pointing at a linear regression: the line WANTS to be predictive

@AOCummies a true friend

@jerrrrryyy_ a loss function is a theological instrument meant to commune with the desires of higher powers

@martin509984 @nectarina12 eat my ass >> suck my dick for both male & female

@NKel26 people who are happy to answer accurately in surveys are likely liberal and wealthy

@forbca @neolibureaucrat https://x.com/christapeterso/status/1285036719717273600?s=20

@dewittd ML has silently revolutionized your life throughout the last 10 years, with computer vision(assisted driving), speech recognition(think Alexa), ranking models(think twitter feed) while all I hear about crispr is some vague rumors about babies in China

@dewittd stuff like GPT-3 and AlphaGo that demonstrate impressive feats of artificial intelligence are just cherry on the sundae scientific monuments

Not being dramatic when I say this weekend has felt like a fever dream or a hallucination. I am suspicious of everything I read https://x.com/quasimondo/status/1285152329449316353

in the future, twitter heads will insult each other w how simply their ideological opponents can be modeled by an AI "lmao I only need a GPT-5 to predict your poasts" https://x.com/aquariusacquah/status/1285126758727999488

@dewittd CV advances have been entirely due to advances in deep learning starting in 2012 at the ImageNet challenge. if autopilot today seems unimpressive to you, you probably don't realize it was non existent in 2010

@dewittd if my daily 5 hours of engagement are any indication the twitter algos are pretty good

@dewittd the biopharma industry is in terminal decline, their r&d investments have negative IRR, and it seems crispr is a PR campaign to convince investors their world isn't ending

@dewittd have you seen footage from the grand challenge? Will answer your question

@dewittd (Those cars were a joke. Not a single one made it to the finish line.)

@dewittd the field went from being intractable to “with enough training data this is solved”. The Tesla model is incredible

@dewittd the only thing Tesla model doesn’t seem to do at the moment is intersections in city driving

@jachaseyoung well NOW I’m thinking it

@dewittd the hardware and software have definitely coevolved — with better hardware, we find algorithms that can scale with more compute power. With new algorithms, we build hardware specialized to run them. Virtuous cycle

@dewittd There was definitely a hype bubble in autonomous tech that left a bitter taste in everybody’s mouth (just another way some of the dumber VCs ruin the interest in good technologies), so agree with you there. Turned out to take longer than folks were promising

@dewittd very much so. Waymo is not even a player on the map anymore as far as I can tell. If *all mammals* can use “software” to mesh together two cameras and create binocular depth perception, I don’t see why the machines need LIDAR

@dewittd I’m not a paid shill I promise :)

@vgr Being “open to new experience” lets people slip instructions into your kernel

@orthonormalist @jeff82874662 @eigenrobot 👀https://x.com/ekarujohnstone/status/1285277410091577344?s=21 https://t.co/XDH0YC0CTw

@orthonormalist @jeff82874662 @eigenrobot no doubt academia is fucked, but there are plenty of phds that will get the job market salivating, this just isn’t one of them

@selentelechia deleting replies bc i think they were somewhat tonedeaf i definitely agree that there is a gut churning cosmic horror vibe when talking to it https://x.com/tszzl/status/1285330054285799427

@selentelechia but low burn cosmic horror which makes it harder to understand

@WhippleMarc @selentelechia gonna have to say i don't think this is in anybody's area of familiarity tbqh lol even the NLP scientists involved have no more say of whether or not its intelligent/conscious/thinking than you do

@fthemt @selentelechia yeah that's exactly how i feel. suspicious of all the text i read

@fthemt @selentelechia i hope someone manages to make this phenomenon legible at the very least will make for some interesting scifi horror

@fthemt same, i'm just like "dammit i fucked up the prompt"

@fthemt @selentelechia https://en.wikipedia.org/wiki/The_Book_of_Sand

@sonyasupposedly @gaitanalyst every religion requires its followers to take purchase in at least one highly ridiculous claim and this one is mine i.e i have a rock solid belief that our civilization will find advancements in the tech tree in the right order & at the right times to continue its gameplay safely

@Noahpinion kinda fascinated always by the internal logic of ideologies i've never even heard of

@Noahpinion but ya you can find similar stuff in hindu nat forums

@selentelechia @fthemt ya the narrator considers that and then reasons it might create endless smoke and engulf the world

@Noahpinion i might just be being naive abt history but i think the cambrian explosion of weird ethnonat ideologies has more to do with the information age than the industrial revolution

@Noahpinion definitely possible but you will have to explain to me why india's industrialization started a few decades ago but they became hindunats in the 2010s

@Noahpinion and why it happened at the same time as the rise of the most race-hawkish arm of the CCP, and also the rise of all sorts of hitherto dead ideologies in the US

@HillaryReplyGuy @Noahpinion im gonna ask this weirdly, but why is the online memeset of BJP politics more powerful than that of the secular nehru-gandhi dynasty?

@HillaryReplyGuy @Noahpinion is it that liberalism is an "unnatural" ideology and can't be sustained when media isn't fully controlled by the elites? is it that liberals across the world are systematically bad at Online?

@arnoblalam @Noahpinion gandhi was deified upon his death, and his shooter's name demonized in history. so of course hindu nationalism isn't a new force at all, but what led to it finally collapsing the nehru gandhi dynasty that had been in power for so long?

@selentelechia @fthemt he leaves it somewhere deep in a library "a leaf is best hidden in a forest"

@arnoblalam @Noahpinion the BJP does plenty of paying its voters off, that's just how politics is done in india (and i would argue across the democratic world)

@arnoblalam @Noahpinion we call it welfare, they call it "free bags of rice if you're in a disadvantaged caste"

@yustar421 @Noahpinion most things boil down to this

@noahopinion yup he's an unhinged asshole

@noahopinion agreed

@noahopinion think its kinda weird that twitter just played along with his shego thing like it was funny/cute

@Noahpinion @yustar421 yeah i think the news anchors were openly discussing this at the time

@ne0agent1c theres a lot of settings fiddling you need to do before it starts being impressive my first thought was that it was very dumb and my reaction slowly turned into complete amazement

@alth0u its just not incentivized since you don't have to maintain research code bases for a long time

@alth0u yeah research ML is so slow and so suboptimal but you also generally get to work with cooler shit

@alth0u Google not gonna let you play with GANs

@alth0u my friends advice to me when I published a paper: “don’t clean up the code, it’ll make it harder for them to find bugs”

@ne0agent1c except the thing is nobody really gets converted by logic, but by aesthetics

@ne0agent1c so first, are you playing the gpt2 version or gpt3? and also, ive heard the ai dungeon can be stunted vs the raw api

@ne0agent1c yup thats true the AI will have manifest somehow in reality before it has aesthetic flair tho build monuments to itself and such

interesting concept, but AIs have been sandbagging for a while now. AlphaGo would develop a 1 stone advantage and play extremely conservatively after that — while a human might enjoy the flair of finishing with way more stones, the AI doesn’t care https://x.com/ESYudkowsky/status/1285333002252247040

Here’s some more evidence of GPT3 clearly sandbagging https://x.com/nicklovescode/status/1284050958977130497?s=21 https://x.com/nickcammarata/status/1284050958977130497

@ESYudkowsky https://x.com/nicklovescode/status/1284050958977130497?s=21 https://x.com/nickcammarata/status/1284050958977130497

@shlevy being religious just seems more adaptive than being irreligious honestly

@Noahpinion college indoctrination is obv real in many cases but it has nothing to do with liberal arts professors — the students radicalize each other

@PereGrimmer subtweeting gpt I see

@AOCummies a champion of civil rights

it learns novel high level grammars for programming based on just a handful of examples, in a way that demonstrates that is has picked up inference-time zero shot learning capabilities of tasks not in its train set that no other model we've seen thus far has come close to doing.

it can pick up mood and tone and character and extrapolate these things like nobody's business. it'll even start complex arguments but will inevitably start gaslighting you with fake information and bad reasoning, flawlessly weaved in, (sounds like most ppl on the internet,,,)

it's clear to me personally that the robot *in many situations* is operating somewhere in the human range of intelligence- such a level of intelligence is a surprising result, seeing as its "brain" is holding 1000x less data than the most basic approximation of the human brain

GPT-3 seems to have skipped a bunch of stages on the evolutionary ladder and as such feels extremely uncanny valley to us. it's like a precocious child with a massive breadth of knowledge that often loses the plot b/c it's learned everything through reading books

animal consciousness probably started off as little more than real time processing of sensor fusion, then slowly climbed up the cognitive tech tree to emotion for relationship purposes, spatial reasoning for advanced planning, abstract reasoning for language, self conception etc

all in all i'm incredibly surprised that we've gotten to this level just 10 years after the birth of modern "deep learning", and the trajectory of the GPT models tells me that we haven't even reached the terminus of this approach, so long as we can scrounge up more compute

we have "consciousness" on an honeybee brain and "high reasoning" on a mouselike brain. The question going forward IMHO is, how to leverage the unsupervised learning of GPT3 with the reinforcement learning of game AIs and create agents that are both *conscious* and *intelligent*

But watching starcraft agents play, it's clear that these have not learned a level of abstract reasoning anywhere close to where GPT-3 is at. Their "brains" are also much smaller, (1000 times smaller than GPT3) (excellent related @gwern article here:https://www.gwern.net/Tool-AI)

but also by the same definition, we've already built nearly conscious bots. For example, you can watch small neural nets play Atari games or leviathans playing Starcraft. Game-playing AI (reinforcement learning) has a real-time OODA loop baked into its core - action and reaction

while such a level of generalization is proof to me that it's intelligent, it's clearly not *conscious* in that there's no OODA loop for it to interact with its surroundings, act, reassess, fight to survive, has rudimentary self-conception(a panda is conscious by this definition)

it seems maybe the Attention Transformer itself is that mystical architecture that was needed to make the jump tl;dr gpt3 is obscenely smart but not conscious. to achieve both I think we will need *agentic* language models

for a few years I thought there was something mystical about symbolic reasoning and that it required neural structures unbeknownst to modern DL to get there -- i've strongly updated to think otherwise based on this facebook bot that does math and GPT3 https://ai.facebook.com/blog/using-neural-networks-to-solve-advanced-mathematics-equations/

@ProfitProfound sometimes u just gotta poast

@rudeloux google can’t write narratives for me as far as I can tell

@rudeloux It’s not just google though. It can learn tasks that it has not seen in its train set

@Noahpinion definitely has already been tapped right? russian misinformation campaigns, weaponized memes, space age propaganda etc

@Noahpinion it's just the beginning though, we are about to dial it up to 11 when countries start weaponizing the natural language bots

@Noahpinion there's a neat scifi horror story about an AI demon that inhabits a guy's computer and subtly shifts all his input and output to goad him into certain ideologies of course, facebook, twitter, youtube have already created a bad approximation of this

@Noahpinion anyway Alex Jones was light years ahead of us on this

@charliebsu @Noahpinion quick! build the Foundation!

@charliebsu @Noahpinion our societies become more low trust by the minute. the price of lying is too low. eventually we will build misinformation immune systems fit for the internet age tand balkanize into info zones where we can actually verify truth

@charliebsu @Noahpinion i think the elites are just as susceptible to info pollution as the average person is. in fact they're more likely to hunt for bad info and get caught in local minima b/c they more often engage with ideas that aren't tested in the daily reality of their life at all

@eigenrobot post rationalists

@eigenrobot blue checks

@Willyintheworld yeah that’s precisely how i see it

the AIDungeon phenomenon is incredible. Future generations of AIs will train on the old AIDungeon logs and finally we will be stuck one day communing with godlike intelligences by LARPing as a dungeon crawler

@ChrisCroy lmao how have they managed this?

thank u,,, tech leader

@DukakisDude there are social ills that are outside of the realm of government to tackle

@DukakisDude it’s very very hard to change culture by technocratic tweaking and sugar taxes and whatever else

@browserdotsys also every dan brown book features middle aged classicists that rescue and sleep with 20 something heroines

@s8mb in fairness, you can see the ratios in this thread the crazies are getting destroyed

@elonmusk Elon, what’s your take on GpT3

i rest easy knowing that even if GRRM dies early, GPT-5 will autocomplete his series

@jachaseyoung flaying our brains you say

@jachaseyoung the way these models work, they can compare between different outputs and tell you which of them seems more likely. So yeah, it could probably pick a best ending out of a given set and refine it further and such

@wowholyfucking blackpilled again

@shill_globalist @wowholyfucking growing fatter every day quarantine continues

@jachaseyoung absolutely dude but I will say, I’m grading on a scale for the AI. I would not have been unduly impressed if a human had written the same thing(s)

@alecbzr dammit bro you shouldn’t tag people in posts where others are poking fun at them he blocked me for this — in the future dropping a screenshot is better than a quote tweet