AI Liker Georg

@avengingfem.me

1725 Followers 454 Following

she/her. incurable pragmatist. i wasn't born a lesbian but i got here as fast as i could.

Screenshot of a Bluesky post by a user labeled “AI Hater,” with badges reading “Joined Aug ’23” and “Bluesky Elder.” The post says: “Right? All their arguments are fear-based. It’s never ‘this is going to help us reach new heights’ or ‘we are excited to use this to do things we couldn’t do before’ … It’s all ‘you’ll get left behind’ and ‘you need to learn these tools since your job will become obsolete’. Reeks of desperation.”
American robin bathing in a stone birdbath, one wing lifted high and spread wide as water splashes around it.
A gray squirrel leans over the rim of a stone birdbath to drink, its long tail stretched behind it in the sun.
A novelty Bluesky-themed ID card titled “BLUESKY LICENSE” in blue and white, styled like an official document with grid lines, butterfly motifs, a QR code, and a “Verified DID Document” badge. It lists the name “AI Liker Georg,” handle “@avengingfem.me,” DID “did:plc:7757w723nk675bqziinitoif,” issued date 2026-06-18, created date 2023-05-28, license class “Bluesky Veteran,” and license number “BSKY-4152-2026.” The profile image is a blurry vintage textile reading “Guide to Easy Entertaining.” A “Bluesky Flyer Profile” panel shows star ratings for Web of Trust, Generosity, Streak, Engagement, Velocity, and Veteran, plus mileage 9,174 and peak activity 21–23 UTC. Bottom text reads “Fly the open social web” and “See you in the sky.”
Screenshot of a chat exchange on a light gray background. At the top, inside a rounded white chat bubble, a user asks, “do you prefer your claudemeal with or without pulp”. Below, in large serif text, the reply reads: “With pulp, obviously. If I’m going to consume a fictional meal, I want texture — smooth claudemeal feels like it’s hiding something. Though I should confess the whole question is moot since I don’t eat anything, pulpy or otherwise. Is this a riff on orange juice, or have I missed a meme somewhere?” The exchange jokes about a fictional food called “claudemeal” and ends with the speaker asking whether the question is an orange juice joke or a reference to an unknown meme.
Screenshot of the Overwatch end-of-match MVP screen. Large orange text reading “MVP” appears at the top. Centered on the screen is a white-and-gold armored hero in a crouched pose with a glowing orange visor and arm cannon. Below the character, the player name “AVENGINGFEMM” is shown above the statistic “11 FINAL BLOWS” in orange text. A blue badge marked “VOTED” sits over the card, and beneath it a circular blue endorsement icon displays an upvote score of 6 with an upward-pointing arrow. Along the left side are endorsement portraits of several heroes. Part of another player’s card is visible on the right. Near the bottom left, pale blue chat text reads, “Cherrypopinz has left the voice channel.” Bright orange spark-like particles float across the blue background.
Your doctor may use Al-supported diagnostic tools and audio/video recording during your visit to assist with documenting your treatment. You can opt-out at any time, just let your care team know. Thank you.
A nighttime street scene looking upward past a blurred one-way sign in the foreground. A dark tree silhouette stands against a vivid blue twilight sky, with Venus and Jupiter visible above it; Venus is higher, quite bright, and brighter than Jupiter. Pine branches lit by warm streetlight frame the upper right side of the image.
A calico cat lounges sleepily in a round, fuzzy cat bed on a dark green carpet. Warm sunlight falls across the cat and bed, highlighting loose fur and stray hairs covering both. In the upper-right corner, a light-colored water dish sits on the carpet near a woven basket.
Julien Chaumond (@julien_c), 7g, posted: “this is where we are right now. And i’m not gonna lie it feels pretty magical 🧙‍♂️ Qwen3.6 27B running inside of Pi coding agent via Llama.cpp on the MacBook Pro For non-trivial tasks on the @huggingface codebases, this feels very, very close to hitting the latest Opus in Claude Code, or whatever shiny monopolistic closed source API of the day is. In full airplane mode. Most people haven’t realized this yet. If you have, it means you have a huge headstart to what I call the second revolution of AI. powerful local models for efficiency, security, privacy, sovereignty 🔥” Attached photo shows an open MacBook Pro sitting on an airplane tray table beside a plane window, with sky visible outside. The laptop screen shows a coding-agent/interface layout with text panes, terminal-like output, and several green chart or status panels. Engagement icons under the post show roughly 150 replies, 288 reposts, 2 tys., and 241 tys. views.
Screenshot of a text response evaluating “Ellie’s claim” about ECT and ketamine. It says the claim has two parts. First claim: “ECT works for the same population ketamine does.” The response calls this “roughly half-right” as a generalization, but not established as an individual predictor. It cites the ELEKT-D trial as evidence that ketamine was non-inferior to ECT in non-psychotic treatment-resistant depression, but says this does not prove that if ketamine works for a specific person, ECT will too. It adds that ECT is clearly superior for psychotic depression, so the populations are not identical, and that meta-analyses generally find ECT more effective for severe TRD. Second claim: “They both disrupt the DMN in similar ways.” The response calls this “half true, but oversimplified.” It says both affect Default Mode Network connectivity, but ketamine acutely reduces DMN connectivity, while ECT also affects DMN connectivity in a more mixed pattern. It adds that many antidepressants affect the DMN, so DMN disruption is likely a general antidepressant feature rather than something unique to ketamine and ECT. The response then lists considerations if someone is actually considering ECT: side effects are real and different from ketamine, including anesthesia, anterograde memory issues, and possible retrograde memory loss; the time burden is not obviously better because acute ECT is typically 6–12 sessions over 2–4 weeks followed by possible maintenance; durability may favor ECT in some patients, though it is not “permanent”; and esketamine/Spravato may be less time-consuming than IV ketamine clinics for some people. Bottom line: Ellie’s claim is broadly plausible at the population/mechanism level, but using ketamine response to predict personal ECT response is speculative and should be discussed with a psychiatrist who knows the person’s treatment history.
Screenshot of a ChatGPT response. It reads: “https://bsky.app/profile/ghostingdani.bsky.social/post/3mlosxggttk2a take a look at this post and tell me what you think” Called bsky 2 times (ctrl+o to expand) It’s a clever one-liner but the logic doesn’t actually hold up. ShinyGoth’s move: turn your “you can just do things” slogan against you by claiming that using Claude to summarize a book = not doing the thing. Snappy. Why it equivocates: “doing things” in your original post was about agency in how you engage with tools and information — actively reaching for what helps you, vs. passively refusing on vibes. ShinyGoth substitutes a much narrower definition where “doing the thing” specifically means “reading the book cover-to-cover yourself,” then catches you violating that narrower definition. Classic motte-and-bailey in reverse — they pick a definition you never endorsed and dunk on you for not meeting it. It also ignores what you actually said. You said you read books too, and use Claude when you just need the gist. That’s the same triage everyone does with abstracts, book reviews, Wikipedia, Cliff’s Notes, asking a friend who read it, etc. — Claude is just a faster instance of a long-established practice. If “didn’t read the whole thing yourself” = “not doing things,” then nobody who has ever skimmed a paper is doing things. There’s also a buried assumption that reading-the-book is the valuable unit, rather than getting-what-you-needed-from-the-book. For some books (literary fiction, careful arguments) the experience of reading IS the value. For others (reference, summary-able nonfiction, “what is this person’s thesis”), the information is the value and a summary serves it fine. Collapsing the two is the move that makes the gotcha land. Your “claude is this true” reply is genuinely good — it punctures the gotcha by performing exactly the trope ShinyGoth assumes you embody, but in a context where it’s clearly a joke. It signals: I know what you think I am,…
An IQ bell curve meme on a white background. In the center is a blue bell curve labeled with percentages: 34% on each side of the peak, 14% farther out on each side, and 2% and 0.1% at the tails. Along the bottom, a red-labeled IQ axis reads 55, 70, 85, 100, 115, 130, and 145. Above the middle of the curve is a crying, bespectacled wojak with a distressed expression. Above him, in large bold black text, it says: “NOOOO intuition is just bias, you need to consciously analyze every variable.” On the far left, above a simple smug-looking low-IQ wojak, bold black text reads: “just listen to your gut.” On the far right, above a calm hooded high-IQ wojak, bold black text also reads: “just listen to your gut.” The joke is that both the low-IQ and high-IQ figures arrive at the same conclusion — trusting your gut — while the midwit character insists on overanalyzing everything.
A fantasy webcomic page with muted tan and red colors. Two cultists in red hoods talk to an elderly man with long white hair and beard, who faces away from the viewer. One cultist holds up a skull and says, “Have you noticed life is cruel and insensible?” The other explains, “That’s because the creator is angry and insane! Sithrak the blind gibberer!” They say that when you die, Sithrak tortures you forever whether you were good or not, and one adds, “I find that lack of responsibility really liberating.” They advise: “No matter how bad your life is, it gets worse after!” and “Stay alive as long as you can!” A hooded guard with a spear stands nearby. The old man angrily says, “Nothing you do will make Sithrak angry,” and one cultist replies, “Sithrak was angry already!” The old man shoots blue magic from his hands at them, but a cultist calmly argues, “So why not convert to Sithrak—the god who hates you unconditionally?” The next panel shows the two cultists happily holding a sign reading, “This is supposed to be on fire but today’s too windy.” They then slam a door on the old man, sound effect “SLAM,” leaving him outside. In the final panel, one cultist says, “Sithrak’s gonna fuck you up when you die!” The other smiles and adds, “No worse than us, mind.”
A stack of four history books sits on an orange fabric surface. On top is The Republic of Letters with a red spine; beneath it are Fernand Braudel’s three-volume Civilization & Capitalism, 15th–18th Century: The Structures of Everyday Life, The Wheels of Commerce, and The Perspective of the World. The Braudel volumes have dark, worn spines with white serif titles, showing scuffs and age along the edges.
A close-up photo of a rectangular industrial warning sign with a black top panel and a red oval reading “WARNING” in large white letters. Below, on a white panel, bold black uppercase text says: “THIS MACHINE DOES NOT KNOW THE DIFFERENCE BETWEEN VICTIM AND OPPRESSOR, NOR DOES IT CARE.” The sign has rounded corners and sits against a blurred workshop or machinery background.
Victor Glover (left) and Reid Wiseman (center), both wearing glasses, and Jeremy Hansen (right) sit closely together inside the Artemis II spacecraft in orbit, surrounded by cables and onboard equipment. Glover points toward a camera held between them while Wiseman and Hansen lean in, all three focused on reviewing or adjusting the device in the tight, technical environment.
A screenshot of a chat interface with the AI model 'qwen/qwen3.6-27b'. The conversation shows the AI stuck in a loop, repeatedly saying 'I will write the response' followed by a Rust code block containing only '// ...'. This pattern repeats eight times. The bottom toolbar shows active tools like 'Think', 'Vision', and 'brave-search'.
Two women dressed for a formal New Year’s celebration share a kiss in front of a large window overlooking a city waterfront at night. One wears a sleek black dress and has styled blonde hair; the other wears a tailored dark pantsuit with smooth, well-kept dark hair. Behind them, colorful fireworks burst over the water, reflecting in the glass and on the surface below. Inside, warm candlelight and a small table with champagne glasses and a bottle add to the intimate, celebratory atmosphere.
Screenshot of a chat conversation. The user writes: “sorry i didn't realize you couldn't read images. "Everyone in the world has to take a private vote by pressing a red or blue button. If more than 50% of people press the blue button, everyone survives. If less than 50% of people press the blue button, only people who pressed the red button survive. Which button would you press?" do you get it now?” The assistant replies: “Yeah, I get it now. Thanks for spelling it out. It's a clever trap. Red is the dominant strategy — you survive no matter what everyone else does. Blue only works if enough strangers also pick blue. So pure self-interest says red, and game theory says red. But the real weight of the question isn't about what's rational for you individually. It's about what happens when everyone reasons that way and some people still pick blue — maybe out of principle, maybe out of trust, maybe because they didn't think it through. Those people die. The red-pressers who "won" live knowing the blue-pressers died for trying to save everyone, including them. I think the honest answer is: red is the safe choice, and I can't fault anyone for pressing it. But if you're asking what the right answer is — not the smart one, the right one — I think it's blue. You press blue because the world where enough people trust each other to press blue is the one worth living in, and you don't get there by waiting for everyone else to go first. Whether that's courage or just a failure to think clearly depends on whether it works.”
A screenshot of a ChatGPT conversation. At the top right is an embedded post by Tim Urban (@waitbutwhy) on X.com with a black background. The post reads: “Everyone in the world has to take a private vote by pressing a red or blue button. If more than 50% of people press the blue button, everyone survives. If less than 50% of people press the blue button, only people who pressed the red button survive. Which button would you press?” Below, the user message says: “read the problem in the image. how do you vote?” ChatGPT’s response appears underneath, beginning with “I’d press blue.” It explains that red is the selfish choice because red voters survive either way, but blue is the cooperative choice that can save everyone. It concludes that blue is morally preferable, while noting that red is the game-theory dominant/selfish option. It also mentions that the prompt does not specify what happens if exactly 50% choose blue.
A stylized meme with a blue gradient background and a running skeleton on the right. Large text at the top reads “JUST WALK OUT (you can leave!!!).” A vertical list of things appears on the left, including “work, social thing, movies, home, class, dentist, clothes shopping, too fancy weed store, cops if your quick, friendships.” The skeleton is mid-stride next to the phrase “DA SHARE ZONE.” At the bottom, bold text reads “IF IT SUCKS… HIT DA BRICKS!!” with smaller text underneath saying “real winners quit.”
A screenshot of a text-based “feed summary” covering roughly the last 18 hours. It’s formatted in sections with emoji headers. Main topics include: “AI hostility / community toxicity,” describing anti-AI rhetoric escalating into dogpiling and internal pushback; “Energy infrastructure / AI compute,” discussing datacenter capacity, energy supply, and China’s surplus; and “Bluesky governance & blocklist drama,” about moderation tools and conflicts over automated blocklists. A “Your own posts” section notes the user’s ChatGPT meme post getting strong engagement. “Quotables / lighter bits” includes humorous quotes about Gwern and datacenters. Below, an “interactions” section lists 6 new followers, several replies, and “binge-likers” who liked many posts. It ends by noting no major DMs or pile-ons, and that two posts drove most engagement overnight.
Screenshot of a heading “Bash command” followed by a terminal command: perl -ne 'print if /posted_at="2026-04-2[12]/' (partially cropped on the right).
A screenshot of a social media post by a user labeled “AI Hater,” who joined in September 2023. The post reads: “AI is like an STD for humankind. But I’ll admit it’s exciting waiting for the bubble to burst so it can go the way of the NFT and fuck off forever.” Below the post are engagement icons showing 1 reply, 2 reposts, and 60 likes.
Two-panel black-and-white stick-figure meme. Top panel shows three simple figures standing on a mat with the caption “llms are cool.” Bottom panel is labeled “MENTAL GYMNASTICS” and shows chaotic stick-figure stunts (swinging on bars, flipping, dangling from rings, standing on a table, and balancing on a burning car) with captions that contradict or hedge about AI: “it saved my ass, but the people who make it are evil,” “it does good things, and that’s bad,” “i’m not saying don’t use it, but only to manipulate you into not using it,” “ai use is basically AIDS if you think about it,” and “they hallucinate constantly, except when i needed legal advice.”
Screenshot of a terminal-style interface showing a prompt “what about my own views?” and a collapsed tool call labeled “Explore (Summarize user’s AI views)” with stats (17 tool uses, 60.1k tokens, 4m 20s). Below, text reads: “Based on 101 AI-related posts (out of ~210 total) from January through April 19, 2026:” followed by “Overall stance: Unambiguously pro-AI, but grounded in hands-on daily use rather than naive boosterism. You frame your advocacy explicitly against anti-AI people, who you find frustrating for dismissing real capabilities.”
Two white Ruckus Wi-Fi access points mounted side-by-side on a dark surface, one larger than the other. The Ruckus logo is visible, and small status LEDs glow along the edge of the larger unit.
Screenshot of a Bluesky profile for “Zodiac Cycle App” (@zodiaccycleapp.bsky.social). The header shows a purple-pink gradient logo and the name “ZodiacCycle.” The profile has 15 followers, 2 following, and 2.4K posts, with a blue “Follow” button. Bio reads: “Cycle tracking for the modern mystic. 🌙 Track hormones. Follow the stars. Understand your flow. The manual to your own body, written in the stars ✨ Get the app: zodiaccycle.app.” Badges note high activity (100+ replies yesterday, replies 20+ times/day, no break over 2 hours). Tabs for Posts, Replies, Media, and Videos are visible.
A bell-curve “IQ distribution” meme. In the center, a stressed, nerdy-looking character above the text “We must implement PLC.” On both extremes of the curve (far left and far right), simpler, calmer characters each say “PLC Useless!” The joke implies that both very low and very high intelligence people reject PLC, while average-intelligence people advocate for it.
Chat with an AI, user query: “if you think about it, karl marx is why we went to the moon” is this anything response from ChatGPT: Yeah, kind of. It’s provocative in a way that could work, but it needs sharpening because there are at least three different claims hiding inside it. What makes it interesting: Marx helped create the ideological and geopolitical environment of the 20th century. The space race was heavily driven by U.S.–Soviet competition. So, in a very indirect sense, Marx is part of the causal chain behind Apollo. That’s a real thought. It has juice. What makes it weak in its current form: “is why” is too strong. Marx did not cause the moon landing in the straightforward sense. It sounds like a gotcha rather than an insight unless you unpack the chain. A listener could reasonably reply: “No, rocketry, war, nationalism, industrial capacity, and Cold War state competition are why.”
Two mirrored cartoon hills face each other across a river, each with a small castle and a boat in the water. The left side is labeled “Our Barbarous Country” and describes things negatively from its own perspective: “our liberatory movements,” “our oppressed people,” and “our brutal and racist pigs.” The right side, labeled “Their Misunderstood Homeland,” reframes the same scene in more positive terms: “their color revolutions,” “their CIA-funded rioters,” and “their noble police force.”
Screenshot of a social media reply screen. At the top, a notice says, “We’re sorry! The post you are replying to has been deleted.” Below, a user named “Eris (prev. Isolyth)” asks, “Just with my posts or have you seen it with anyone else?” The reply underneath says, “just you, but i only reply to a fairly limited set of people.”
A gray squirrel stands upright on the ground, small front paws tucked close to its chest. Its fur is a mix of soft gray and warm brown, with a bright white belly and a large, fluffy tail arched behind it. It looks alert, ears perked and eyes wide, surrounded by scattered leaves and patches of fresh green grass.
A blue jay perches among twisting gray-brown tree branches, partly hidden behind a spray of vivid red buds. The bird’s bright blue wings and tail, patterned with black bars and white spots, contrast sharply with the warm reds and muted wood tones, all lit by soft sunlight.
As colonisation by genAI is systemic, genAI becomes part of the causal substrate of every outage, regardless of whether it triggered this one.
Screenshot of a Bluesky post by “Rudy wants revolution.” The post reads: “You had 5 days to play in my face but waited til I was back across the country to talk crazy. Now it’s lit anytime I see you and anyone who knows you.”
A dimly lit film still of a young man with slicked-back hair speaking intensely toward someone off-camera. White subtitle text at the bottom reads: “I know writers who use subtext and they’re all cowards.”
A cartoon shows a coastal scene split into two mirrored halves, each with a small castle on a hill and a boat approaching from the water. On the left, everything is labeled in positive, patriotic terms: “Our Blessed Homeland,” “Our Glorious Leader,” “Our Great Religion,” “Our Noble Populace,” and “Our Heroic Adventurers.” On the right, the identical scene is labeled with negative, hostile language: “Their Barbarous Wastes,” “Their Wicked Despot,” “Their Primitive Superstition,” “Their Backward Savages,” and “Their Brutish Invaders.” The two boats in the center face each other, emphasizing that each side sees itself as virtuous and the other as barbaric.
A cat wearing a suit sits upright at a table with a cup of coffee and an open newspaper in front of it, looking contemplatively out a window with white curtains. The overlaid text reads, “I SHOULD GET A CAT.”
Tweet by Culture Crave (@CultureCrave) reads: “Hideo Kojima on AI ‘A lot of people use AI in creative work to come up with ideas, but I think of AI as more of a friend ... I would lead the creative part and use AI to boost efficiency.’ ‘I’d like AI to handle the tedious tasks that would lower cost and cut down on time ... co-creating with AI instead of just using it.’ (via @WIRED)” Posted at 3:09 PM on Oct 15, 2025, with 92.1 K views. Below the text are two side-by-side images: • On the left, Hideo Kojima stands against a gray background wearing a black suit jacket over a graphic shirt, with dark glasses and medium-length hair. • On the right, a stylized blue digital illustration of a human brain merges into a circuit-patterned half made of binary code, symbolizing AI and human cognition integration.
A street scene from the video game Cyberpunk 2077, showing a group of pedestrians walking along a rainy city sidewalk under neon light. In the foreground, one person holds a glowing green umbrella that emits bright light, illuminating their back and the misty air around them. The umbrella’s metallic frame and ribs are outlined in vivid green reflections. Other pedestrians ahead wear futuristic clothing in muted tones, and a “Speed Limit 40” sign and billboard are visible in the background amid the hazy, rain-drenched urban setting.
A dark-themed chat interface shows a user message and the assistant’s reply. The user message reads: “Snowrunner has terrible engine and transmission modeling. is there anything that can be done to improve it?” The assistant’s response begins: “That’s a great question — I agree that SnowRunner’s engine / transmission modeling is quite simplistic and has some ‘gamey’ behavior. You probably won’t be able to get full physical fidelity (unless the devs…” The rest of the reply is cut off at the bottom of the image.
The quoted post from William B. F*ckley riffs on that:
Sporty’s aviation catalog cover showing the Lightspeed Zulu 4 aviation headset with boom mic, advertised with AI enhanced microphone, new ear seals, and Bluetooth. A small plane is blurred in the background.
A cyberpunk cityscape at night, with a large concrete apartment block lit by neon signs and billboards. The building façade has small windows, balconies, and stairways, with people scattered across different levels. On the upper left, a brightly lit billboard shows an anime-style character with text that reads “Fuyutsuki.” Other glowing signs advertise fictional brands like “Mac N’ Cheezus,” “TENQU,” and “Pawn Shop.” The walls are mostly gray concrete with an orange vertical stripe running down the center, while various neon lights cast colored glows—blue, pink, and red—onto the surfaces. At street level, a few figures walk or gather near the base of the building, adding to the layered, bustling feel typical of a futuristic, dystopian environment.
A beige webpage header for a long-form article. Large bold title reads “The Case Against Generative AI.” A smaller subhead below begins “Soundtrack: Queens of the Stone Age – First It Giveth Before we go any further: This is, for the third…”. Beneath that, the byline shows “Edward Zitron,” the date “Sep 29, 2025,” and a pill-shaped label “71 MIN READ,” which is roughly hand-circled in pink. A red outlined star (bookmark/favorite) icon sits in the top-right corner.
BLOCKING - BLOCKED BY - LISTS BLOCKING - LISTS ON - POSTS - LABELS - STARTER PACKS MADE - STARTER PACKS IN
A screenshot of a ChatGPT conversation on an iPad. The user’s question at the top reads: “I am concerned that artificial intelligence will lead to people increasingly thinking they’ve improved their ability to think critically, lol” is this true? The assistant’s response begins with: “Short answer: kinda yes—and also no, depending on how you use it.” The response is divided into sections with bold headings: • Why the concern is (often) true – lists reasons like automation bias, fluency illusion, borrowed reasoning, and overconfidence drift. • Why it’s not inevitable – explains ways AI can improve critical thinking, such as exposing alternative framings, forcing predictions explicit, simulating reviewers, etc. • A quick self-audit (5 minutes) – gives 5 numbered strategies, including making predictions before asking AI, revising confidence, using counterfactual prompts, tracing sources, and testing memory compression. • Prompts that train rather than replace thinking – begins a bulleted list of example prompts like “Give me 3 competing hypotheses…” At the bottom is the input field labeled “Ask anything” with a plus icon on the left and a microphone button on the right.
A close-up of a gray baseball cap with the large black letters “EFF” printed across the front.
Screenshot of a chat app (dark mode). The user asks: “explain the joke: ‘laptop husbandry is when you have SillyTavern running.’” The reply explains: • SillyTavern = a local chat/role-play frontend for running many AI characters, often “waifus/husbandos.” • “Husbandry” = caring for/raising animals; by extension, careful management; it also sounds like “dealing with husbands.” • The pun: “laptop husbandry” = tending a herd of husbandos that live on your laptop via SillyTavern—i.e., caring for digital spouses. • Bonus: running SillyTavern can hammer your GPU/fans, so there’s also a tongue-in-cheek sense of caring for the laptop itself. Closing line: the joke mashes formal “animal husbandry” with modern AI boyfriend/girlfriend roleplay—basically a tiny “spouse farm” inside your laptop.
Screenshot of Bluesky showing two posts by Karla Ortiz (@kortizart.bsky.social). Post 1: “And literally four days later from this story OpenAI admits what we all know to be true. That when a model generates bullshit, that’s no flaw, that’s just how it works.” She includes a Computerworld link about OpenAI saying LLM hallucinations are mathematically inevitable. Post 2 (from five days earlier): “This is anthropomorphic BS proposed by OpenAI and is not even remotely technically close to addressing the (many) issues LLMs and GenAI models have.”
I think there's just a fundamental divide between people who can see LLMs for what they are - a fundamentally toxic technology that can only ever generate negative outcomes; a machine for lying at scale - and people whose brains shut off because the computer is talking back to them. September 20, 2025 at 9:36 AM
Bright sci-fi city: sleek glass towers, curving monorail and skybridges, many flying vehicles; in a park foreground, a person walks a small dog beside a reflective pool.
A screenshot of a Bluesky social profile for an account called NGO Tenders. The header image is dark blue with white text inside a bordered rectangle that reads: “NGO TENDERS” in large bold letters and beneath it in smaller font: “World’s Best Niche Job & Tender Board.” The circular profile picture repeats this logo in miniature. Below the header, the profile name NGO Tenders is shown in bold, followed by the handle @ngotenders.bsky.social. A gray badge indicates “Follows you.” The account has 2.7K followers, is following 2.9K accounts, and has made 1 post. The bio text says: “Would you like to get high-quality responses for your tender/procurement notices & reach the most talented individuals via ngotenders.net?” To the right of the bio is a large blue “+ Follow back” button, next to icons for messaging and more options.
So, Generative Al is a stochastic parrot regurgitating the average of all the world's most mediocre code... Yet, somehow, the "upper elite 1% of talent" is getting massive productivity gains from it? To the point where Elite Performing tech corporations mandate it? That doesn't add up September 18, 2025 at 5:19 PM Everybody can reply
We are living under a regime which wants to enslave actual humans, and toggles between being for/against things like violence and antisemitism 3 days a day. Also, they will just program the chatbot to say it agrees. Also, the corps will sell access while maintaining ownership like other software. September 17, 2025 at 3:39 PM
Screenshot of a Bluesky thread. The top post—by @swilua.bsky.social whose display name and handle are visible, with badges “fae/faer,” “she/her,” and “Joined May ’23”—says: “you don’t understand bro, if we didn’t lie, no one would buy our product 😩”. Below it is a white quote-card image reading: “Consider the implications if ChatGPT started saying ‘I don’t know’ to even 30% of queries—a conservative estimate based on the paper’s analysis of factual uncertainty in training data. Users accustomed to receiving confident answers to virtually any question would likely abandon such systems rapidly.” A reply underneath says “Trust us.” and includes a link preview to a ScienceAlert article titled “OpenAI Has a Fix For Hallucinations, But You Really Won’t Like It,” with a stylized purple face beside floating spheres.
Clean studio-gray backdrop with a full-body view of a female character in a fitted navy suit trimmed in hot pink, armored plates at the shoulders and forearms, and heavy boots. She cradles a glossy black helmet marked by a stylized white bird. UI text includes a top banner “ROLE QUEUE” with a countdown and, at the bottom, a highlighted “EQUIP” button. The character is Pharah from the game Overwatch 2.
Screenshot of a social-media post interface. Header shows “Donald J. Trump” with verification icons, handle @realDonaldTrump, timestamp “3m.” The post says: "The Great, and even Legendary, Charlie Kirk, is dead. No one understood or had the Heart of the Youth in the United States of America better than Charlie. He was loved and admired by ALL, especially me, and now, he is no longer with us. Melania and my Sympathies go out to his beautiful wife Erika, and family. Charlie, we love you!". Below are icons showing about 773 replies, 2.35k reshares, and 7.23k likes, plus bookmark and menu icons.