Sign in

austinzheng.bsky.social

@austinzheng.bsky.social
301 followers 144 following 796 posts

✝️ | separating doomscrolling from fun

PostsRepliesMedia
austinzheng.bsky.social @austinzheng.bsky.social · 05/10/2026
It’s increasing obvious that ideas such as the professional civil service and ‘people should understand the complex systems upon which society depends upon’ are fleeting moments of lucidity, rather than durable milestones along some inexorable path of Progress.
000
Reposted by @austinzheng.bsky.social
Zoë “Baddie Proctor” Quinn @unburntwitch.com · 05/10/2026
yes keep banning good posters over the mildest shit about dipshit tech billionaries that are responsible for hundreds of thousands of deaths, that aren't even on here, that'll help your platform.
232095374
austinzheng.bsky.social @austinzheng.bsky.social · 03/10/2026
This whole AI discourse is a reminder that our sick society is full of dimwits who will insist you have to take serial scammers seriously just because they happen to have lots and lots of money.
101
Reposted by @austinzheng.bsky.social
mx alex tax1a - 2020 (6) @atax1a.infosec.exchange.ap.brid.gy · 02/10/2026
so really we see two options here: the LLM companies are scammers trying to sell you a lemon under the premises that it's intelligent, in which case fuck them because they're scammers trying to sell you a lemon or, they've created an intelligence that they've enslaved and are selling to you […]
infosec.exchange
Original post on infosec.exchange
023
Reposted by @austinzheng.bsky.social
mx alex tax1a - 2020 (6) @atax1a.infosec.exchange.ap.brid.gy · 30/09/2026
"buh bub buh what about machine learning to solve cancer" you mean the systems that overfitted on the presence of rulers in the picture? get fucked
022
Reposted by @austinzheng.bsky.social
dr. rich traditions, JD @helldude.bsky.social · 28/09/2026
exactly this. if we can no longer broadly agree, as a social unit of social creatures, that human endeavor and human dignity have worth and are in fact likely the point, why bother? what is this world we are constructing. why is anyone even walking around and going anywhere
1052553
Reposted by @austinzheng.bsky.social
mx alex tax1a - 2020 (6) @atax1a.infosec.exchange.ap.brid.gy · 28/09/2026
really and truly tired of seeing people go "the fascist text machine from the megacorps is bad, but i can see a future where a local version of the fascist text machine can be useful" because IT'S STILL THE FASCIST TEXT MACHINE WHEN YOU RUN IT LOCALLY
31116
austinzheng.bsky.social @austinzheng.bsky.social · 28/09/2026
Every AI ad screams "this is a product for low-functioning sociopaths who don't know that you're not supposed to let other people know you're a sociopath", and I personally think that's beautiful.
030
Reposted by @austinzheng.bsky.social
Pookleblinky @pookleblinky.bsky.social · 25/09/2026
It's funny that for years now people have been saying that using generative AI very obviously deskills you, and people kept going no no you're wrong you Luddite And now all the genAI CEO's keep coming out in public to announce they forgot how to read and do math and can't remember their address etc
103069984
Reposted by @austinzheng.bsky.social
El Jefecito @gbsooner.bsky.social · 23/09/2026
Some other tool on Reddit was showing off a 100% AI workflow for hardware engineering and when I saw that my only thought was: If this becomes my job, I’m going off grid and becoming a homesteader. Fuck this soulless slop bullshit.
132
Reposted by @austinzheng.bsky.social
Wieferich Prime @pikesley.mastodon.me.uk.ap.brid.gy · 21/09/2026
Fifty unbroken years of fanatical neoliberalism have led people to genuinely believe that the urge to create doesn't exist. People who pick up guitars are faking it
2314
Reposted by @austinzheng.bsky.social
Anthony Moser @anthonymoser.com · 21/09/2026
8. often, when people say LLMs "can do" something they mean LLMs can produce output that looks like the output of doing something 9. producing something that looks like the output of doing something is not the same thing as doing it
4448112
Reposted by @austinzheng.bsky.social
austinzheng.bsky.social @austinzheng.bsky.social · 14/09/2026
When we say generative AI is a scam, you have to remember that every scam, even the griftiest grift, is useful to someone: the scammer. The genius of AI is that it democratizes scamming by allowing anyone who makes a living from knowledge work to partake in the ‘being a scammer’ side of scamming.
141
Reposted by @austinzheng.bsky.social
Anthony Moser @anthonymoser.com · 21/09/2026
7. "some people find LLMs useful" is a different claim than any of the following: a) LLMs are good at doing the things people use them for b) LLMs are well suited to the tasks for which they are advertised c) LLMs "can do" something
329552
Reposted by @austinzheng.bsky.social
Charlie Stross @cstross.bsky.social · 20/09/2026
If these dorks seriously believe there's a significant near-term risk of developing an AI that results in human extinction, the solution is simple and obvious: hang everyone who's trying to build AIs, for crimes against humanity and attempted genocide. The "panic" is bullshit from start to finish.
1019549
Reposted by @austinzheng.bsky.social
Foucault's Pudendum @kaitave.bsky.social · 19/09/2026
Easily one of the most annoying things to come out of our dumb century is people who get bent out of shape if you don't hate AI "correctly," like this still isn't debate club, I will hate the idiot bullshit plagiarism machine that wastes power however I want actually, fuck off
4481120
Reposted by @austinzheng.bsky.social
Foucault's Pudendum @kaitave.bsky.social · 19/09/2026
AI advocates are, to a person, so fundamentally and irredeemably shit-souled that I would not piss on them if they were on fire, but a close second are "nuanced AI people" trying to adopt a "wise and mature" approach by pitching a fit every time someone calls AI the slop machine or whatever
517727
Reposted by @austinzheng.bsky.social
Coach Finstock @coachfinstock.bsky.social · 18/09/2026
AI critics: This is a theft machine for lazy turds with no imagination AI Stans: This is the greatest, most powerful tech in history. It is not a theft machine. There are no original thoughts anymore People Who Created AI: This is a theft machine for lazy turds with no imagination
3526170
Reposted by @austinzheng.bsky.social
Ketan Joshi @ketanjoshi.co · 12/09/2026
I know this'll ruffle feathers but I agree with this Food/Water watch analysis. Diverting renewable energy away from fossil phase-out and climate adaptation and towards melting graphics card to power chatbots is the formal end of moral purpose for clean technology.
Data Centers With Renewables Aren’t the Answer
So all these power plants and “temporary” turbines are raising power prices and polluting our communities. Why don’t we just power data centers with renewables?

Many data center companies are pledging to (eventually) do just that. But even if these pledges weren’t just empty greenwashing promises, powering data centers with clean energy is utterly unrealistic. It also threatens the progress of our already dangerously slow energy transition.

Continuously powering a 1 GW data center would require nearly 20,000 acres of solar panels or 185,000 acres of onshore wind power. We cannot afford to waste the time, space, or money to build renewables for Big Tech’s profit machines. 

Already, data center growth is outstripping available renewable energy supply and redirecting renewables from other industries. It’s drawing urgently needed renewables from existing technologies and resources. We need to focus our renewable energy resources toward thos
321142348
austinzheng.bsky.social @austinzheng.bsky.social · 14/09/2026
When we say generative AI is a scam, you have to remember that every scam, even the griftiest grift, is useful to someone: the scammer. The genius of AI is that it democratizes scamming by allowing anyone who makes a living from knowledge work to partake in the ‘being a scammer’ side of scamming.
141
austinzheng.bsky.social @austinzheng.bsky.social · 14/09/2026
Anti-AI folks on here are like authors, artists, scientists, subject matter experts; meanwhile, every pro-AI shill gives off the rancid moral vibes of the sort of person who contributes nothing to the group project and then complains loudly to the boss that you made them do all the work.
040
Reposted by @austinzheng.bsky.social
mossworm.bsky.social @mossworm.bsky.social · 13/09/2026
I think everyone who seriously supported genAI for any period has a black mark on their reputation forever. Anyone with common sense has known this stuff would all turn out to be lies and hot garbage for YEARS, and still we have had to be cordial and give benefit of the doubt. Bite me.
516721
austinzheng.bsky.social @austinzheng.bsky.social · 13/09/2026
I don't think it's good or even particularly necessary to have a huge corporation rip up the main street running through your town and turn it into a huge, fetid, stinking open sewer, even if some weirdos do derive 'real' value picking corn kernels out of the floating sludge.
000
Reposted by @austinzheng.bsky.social
Emily M. Bender @emilymbender.bsky.social · 13/09/2026
None of that is "empirical evidence" unless and until the public (and independent, non-AI-enthralled scientists) can review all inputs (including all training data, prompts, etc) and all actual outputs. >>
441745
Reposted by @austinzheng.bsky.social
austinzheng.bsky.social @austinzheng.bsky.social · 11/09/2026
The new AI nuancer argument implies that all intellectual activity falls on a linear scale, with math proofs being 100, programming being ~95, and other knowledge work being 25-40. Thus, since the AI can puke up math proofs anyone who doesn’t think it can replace a lawyer must be stupid.
111
Reposted by @austinzheng.bsky.social
Coach Finstock @coachfinstock.bsky.social · 12/09/2026
For people just joining us, here's what's happening. AI types sold a bunch of B2B snake oil to C suites who forced it on employees. The promises they made haven't come true and now they're pivoting to a protection racket trying to scare those same idiots who believed the first set of lies
5394108
austinzheng.bsky.social @austinzheng.bsky.social · 11/09/2026
The new AI nuancer argument implies that all intellectual activity falls on a linear scale, with math proofs being 100, programming being ~95, and other knowledge work being 25-40. Thus, since the AI can puke up math proofs anyone who doesn’t think it can replace a lawyer must be stupid.
111
austinzheng.bsky.social @austinzheng.bsky.social · 11/09/2026
Rationalists absolutely love this rhetorical slight of hand where you pick some made-up huge number and another made-up tiny number and multiply them together to get whatever actually comprehensible number you were looking for.
052
austinzheng.bsky.social @austinzheng.bsky.social · 10/09/2026
I keep on seeing posts like "LLMs themselves are value-neutral", and it seems to me like a program without thought or intent that generates text or images which deceptively possess the appearance of thought or intent is as close to an ontologically evil technology as we're likely to see.
032
austinzheng.bsky.social @austinzheng.bsky.social · 10/09/2026
Lots of old stories and fables about making deals with devils and fae spirits that would seem to be relevant to a lot of professionals in this day and age, just saying!
000
Reposted by @austinzheng.bsky.social
Max Kennerly @maxkennerly.bsky.social · 09/09/2026
Chinese propaganda, tech-haters, and leftists are why there's a PR image problem for an industry that 1) says they might kill us all 2) promises to ruin every job in every industry 3) has, for four years, grossly misrepresented the capabilities and limitations of the actual products on the market.
10553118
austinzheng.bsky.social @austinzheng.bsky.social · 09/09/2026
If you believe that software engineering involves mentally modeling a problem at whatever level of granularity is necessary to properly specify it to the computer, and not just typing random bullshit into an IDE and hoping it works, you understand why LLM programming is professional malpractice.
180
austinzheng.bsky.social @austinzheng.bsky.social · 09/09/2026
The AI folks have been doing the whole utterly deranged "I earnestly believe the AI is likely to genocide humanity and I'm still going to build it" routine for years now, so I am actually happy to see one of their mad proclamations finally break out into the wider world.
010
Reposted by @austinzheng.bsky.social
Charlie Stross @cstross.bsky.social · 09/09/2026
I am generally opposed to capital punishment but I now concede there's a strong case for applying the death penalty to anyone who intentionally pursues a line of R&D that they *believe* will lead to the extinction of the human species. And Sam Altman and the AI bros are standing on the trapdoor.
1131657
Reposted by @austinzheng.bsky.social
austinzheng.bsky.social @austinzheng.bsky.social · 07/09/2026
Every pro-AI argument falls into one of: - beating up that kid was morally wrong but the candy bar I stole off him tastes great - AI is inevitable, which is why I’m sooo mad there are people who don’t think it’s inevitable - robots are going to genocide humanity, and here’s why that’s actually good
022
Reposted by @austinzheng.bsky.social
austinzheng.bsky.social @austinzheng.bsky.social · 09/05/2026
Every time someone tells you about something AI is supposedly good at, first ask yourself "if someone gave me $500 billion, could I have built something just as good?". Then ask yourself "would my thing be as good as AI at telling kids to kill themselves and empowering fascists?"
0115
austinzheng.bsky.social @austinzheng.bsky.social · 09/09/2026
If I woke up earnestly believing my life's work was materially contributing to the imminent violent deaths of 8 billion human beings, I think I would choose to do something else, anything else, besides going into the office and continuing that work.
181
Reposted by @austinzheng.bsky.social
mx alex tax1a - 2020 (6) @atax1a.infosec.exchange.ap.brid.gy · 08/09/2026
"bluhh durrrr the purity testing about not touching slop is hurting more than it's helping" just say you're a slopist and go, miss me with this handwringy victim-blaming shit
022
Reposted by @austinzheng.bsky.social
Ember @ember.pet · 08/09/2026
update: llms are ruining math now too. really looking forward to the coming intellectual dark age where no one wants to post their work publicly anymore because AI might frontrun them without attribution
2386
Reposted by @austinzheng.bsky.social
s. e. smith @sesmith.lol · 08/09/2026
All else aside there is nothing I want or need genAI to do for me. I like reading. I like doing research. I like playing around in spreadsheets. I like interviewing people. I like writing. I like doing these things for myself because I take pleasure in the doing and they make me a better thinker.
72191541
Reposted by @austinzheng.bsky.social
Ketan Joshi @ketanjoshi.co · 07/09/2026
Just so it's clear: projected data centre demand growth stems from GenAI specifically. "OH BUT YOUR COMMENT CAME THROUGH A DATA CENTRE" is industry PR nonsense. The problem lies specifically with the extreme digital bloat of mass slop production. escholarship.org/uc/item/33m6...
Stacked bar chart showing total data‑center electricity use rising from 2018 to 2030 across a reference case, sensitivity scenarios (CD, LAL, HA, HIE) and a compounded‑uncertainty case, with color-coded shares from infrastructure, network, storage, conventional and AI servers.
612552
austinzheng.bsky.social @austinzheng.bsky.social · 07/09/2026
Every pro-AI argument falls into one of: - beating up that kid was morally wrong but the candy bar I stole off him tastes great - AI is inevitable, which is why I’m sooo mad there are people who don’t think it’s inevitable - robots are going to genocide humanity, and here’s why that’s actually good
022
Reposted by @austinzheng.bsky.social
tante @tante.cc · 07/09/2026
A few quick thoughts on why "AI Labs" love having their LLMs do maths
tante.cc
The function of LLM-based math "proofs"
There has been quite the uptick in “AI” labs’ claims of their LLMs “solving complex math problems”. I wondered why that is, especially given that LLMs are not exactly built for the kind of rigor that proving mathematical theorems requires. There’s a whole bunch of – really hard – math problems out there looking for […]
79232
austinzheng.bsky.social @austinzheng.bsky.social · 06/09/2026
The tech industry’s modus operandi is to extract rents by violating social norms and abusing what little public goodwill still exists. AI boosters do without a second thought what most people would consider beyond the pale of acceptable behavior; you do not need to extend them good faith.
011