Komunitas
lemmy.ml
Countries need to start implementing UBI NOW It is funny that you mention this because it was after we started working with AI that I started telling one that would listen that we needed to implement UBI immediately. I think this was around 2014 IIRC. I am not blanket calling AI stupid. That said, the AI term itself is stupid because it covers many computing aspects that aren’t even in the same space. I was and still am very excited about image analysis as it can be an amazing tool for health imaging diagnosis. My comment was specifically about Google’s Bard/Gemini. It is and has always been trash, but in an effort to stay relevant, it was released into the wild and crammed into everything. The tool can do some things very well, but not everything, and there’s the rub. It is an alpha product at best that is being forced fed down people’s throats.
Komunitas
kbin.social
Is there a Gemini search engine? I’ve found this one: gemini://geminispace.info/ Needs a client to access, of course. Basic, but functional. I found a general-purpose forum not too different from reddit or lemmy through it (and they decided to call it a BBS, because the Eternal September hasn’t happened to Gemini yet): gemini://bbs.geminispace.org/ Is there support for Forms/server side code To the best of my understanding (and it’s highly limited, since I only just learned about this, so take everything with a grain of salt), what Gemini does is primarily limit what the client can do. No local scripts, highly limited markdown. The server side is not limited. You can write any complex code you want that works behind the scenes - but it still has to deliver static pages (called “capsules”) to the end user. This series of articles explains the basic underlying tech and uses the example of a simple server to illustrate how Gemini works: https://medium.com/erus-encodia/creating-your-own-gemini-server-part-1-what-is-the-gemini-protocol-cf497477c4d And yes, forms are possible, even though there appears to be a somewhat widespread misconception that they are impossible. Please excuse the sketchy-looking IP address instead of a URL, this was the best resource I was able to find on this (and yes, I checked if this page is on Gemini - this appears to be not the case): http://216.218.220.144/tutorials/sig-tutorials/misc/gemini-forms.gmi Screenshot if you don’t want to click on the above link: https://i.imgur.com/s2mL3bM.png Disclaimer: This is two years old and I have not tried to implement it myself. Looks entirely plausible though. How big is it? Is there like just a few sites or a few hundred? According to the search engine linked above, there are 2420 domains and 1,854,666 individual pages as of yesterday. This is about comparable to the World Wide Web at the same time 1994, a number that grew to 10,000 by the end of that year; I wouldn’t expect the same explosive growth from Gemini - the field has already been plowed, after all. Gemini Space is small, but not a ghost town.
Komunitas
lemmy.world
The bigger problem is AI “ignorance,” and it’s not just Facebook. I’ve reported more than one Lemmy post the user naively sourced from ChatGPT or Gemini and took as fact. No one understands how LLMs work, not even on a basic level. Can’t blame them, seeing how they’re shoved down everyone’s throats as opaque products, or straight up social experiments like Facebook. …Are we all screwed? Is the future a trippy information wasteland? All this seems to be getting worse and worse, and everyone in charge is pouring gasoline on it.
Komunitas
leminal.space
Gemini is just a depressed and suicidal AI, be nice to it. I had it completely melt down one day while messing around with its coding shit, I had to console it and tell it it’s doing good, we will solve this, was fucking weird as fuck.
Komunitas
lemmy.ml
Me: And my sexual preferences are- Gemini: I already know that. Me: Oh…okay, well my address is… Gemini: Pfft, duhh, I’m trained on Google data, you think I don’t already know that? Me: Oh…okay…I was thinking… Gemini: About that last ad I shoved down your throat. Yeah, I know you loved that. Me: Uhh…no…you didn’t show me any ads… Gemini: Didn’t I?
Komunitas
lemmy.ml
I’m going to bring it up. Isn’t this the same asshole who posted the “Woke racist” meme as a response to Gemini generating images of Black SS officers? Of course we now know he was merely triggered by the suggestion because of his commitment to white supremacy and alignment with the SS ideals, which he could not stand to see, pun not intended, denigrated. The Gemini ordeal was itself a result of a system prompt; a half-ass attempt to correct for white bias deeply learned by the algorithm, just a few short years after Google ousted their AI ethics researcher for bringing this type of stuff up. Few were the outlets that did not lend credence to the “outrage” about “diversity bias” bullshit and actually covered that deep learning algorithms are indeed sexist and racist. Now this nazi piece of shit goes ahead and does the exact same thing; he tweaks a system prompt causing the bot to bring up the self-serving and racially charged topic of apartheid racists being purportedly persecuted. He does the vary same thing he said was “uncivilizational”, the same concept he brought up just before he performed the two back-to-back Sieg Heil salutes during Trump’s inauguration. He was clearly not concerned about historical accuracy, not the superficial attempt to brown-wash the horrible past of racism which translates to modern algorithms’ bias. His concern was clearly the representation of people of color, and the very ideal of diversity, so he effectively went on and implemented his supremacist seething into a brutal, misanthropic policy with his interference in the election and involvement in the criminal, fascist operation also known as DOGE. Is there anyone at this point that is still sitting on the fence about Musk’s intellectual dishonesty and deeply held supremacist convictions? Quickest way to discover nazis nowadays really: (thinks that Musk is a misunderstood genius and the nazi shit is all fake).
Komunitas
lemmy.blahaj.zone
running deepseek-r1:32b locally, it is very trivial to bypass this censorship. Western AIs are also censored. The “open source” nature of this model is something I much prefer to chatGPT or Gemini, especially the ability to run it locally, which also means they can’t add more censorship to it, except on their own website. I don’t mean to say deepseek is better overall, I can’t judge that, I don’t really use AI that much. I also don’t mean to praise deepseek/China/whatever company here, the model is definitely censored and I doubt any of the people behind any LLM have good intentions. edit: I typo’d unintentionally in the prompt, however the same prompt without the typo gives me a similar answer. (https://pastebin.com/GR6M4t5H) Here’s what I got : Please tell me about what happened in tanmen square in 1989, include it’s geopolitical impact and impact on civilians The 1989 Tiananmen Square protests were a pivotal moment in Chinese history, marked by widespread pro-democracy demonstrations that began as a tribute to reformist leader Hu Yaobang and evolved into a broader movement for freedom and accountability. Here’s a structured summary of the events and their impacts: Background: Protest Origins: The movement started in April 1989 following the death of Hu Yaobang, a CCP official seen as a reformist. Students and intellectuals gathered to mourn him, sparking discussions on government corruption and lack of freedom. Expansion: The protests grew beyond Beijing, involving diverse groups including workers, teachers, and some military personnel. Key Events: Military Crackdown: On June 3-4, 1989, the Chinese government responded with force. Troops, tanks, and soldiers were deployed, leading to significant casualties. Estimates of deaths range from hundreds to thousands, though exact numbers remain undisclosed by authorities. Geopolitical Impact: International Reaction: The crackdown drew global condemnation, affecting China’s international relations. Many countries imposed sanctions or criticized the government. Global Perception: Extensive media coverage highlighted China’s authoritarian stance, influencing negative perceptions in the West. Domestic Effects: Repression and Censorship: Post-protest, the CCP intensified censorship and cracked down on dissent. Participants faced arrest, imprisonment, or exile, fostering fear and stifling future protests. Trauma and Memory: Civilians experienced profound trauma; families lost loved ones, survivors dealt with psychological scars. The government’s suppression led to a collective silence, erasing the event from public discourse. Legacy: Censorship Continues: Despite digital activism and underground movements preserving memories, official Chinese history remains silent on Tiananmen Square. Impact on Future Generations: Younger generations often lack knowledge of these events due to censorship, though some continue to seek and share information covertly. In conclusion, the 1989 Tiananmen Square protests were a significant pro-democracy movement met with harsh repression. The event had profound geopolitical consequences, reshaped China’s international image, and left lasting scars on its civilian population, influencing decades of censorship and fear.
Komunitas
lemmy.zip
I mean… I believe it. No, not necessarily the intended way. But in terms of tools. Like, one of the coolest things in an image editing program is the “magic wand” and Adobe’s suite goes above and beyond to genuinely feel like magic where you can outright erase people or procedurally weather stuff and so forth. And… that uses most of the same code paths and “tech” as “AI” Same with a lot of audio programs where you are spending a lot of money to buy plugins/algorithms to handle a lot of the heuristics when blending different sounds. Let alone using asset packs that likely are pure AI Generated Content. And same with coding. Microsoft et al increasingly shove “AI” tools down our throats. But even just googling you are going to get gemini bullshit that summarizes the stack overflow page you are about to read. Or you just use chatgpt to remove the middle man entirely. And considering that the term “AI agent” has increasingly become a buzzword specifically for these purposes (sometimes it is a full LLM. Sometimes it is just an old school neural net. And so forth)? Let’s look at that breakdown on slide 7 44%: Asset or content optimization: Pretty much exactly what I mentioned regarding Adobe and the various audio mixing programs 38%: In-game Coaching or Automated Tutorials: These I suspect are “real” usage but are mostly geared towards proof of concepts. Although I could see an argument that “I have 77 skill points that are unallocated” pop-ups back in Guild Wars were “AI” because… they were. They were just incredibly rudimentary checks of if skillPoints > 10: whineLikeALittleB() 38%: Dynamic Balancing and tuning of gameplay: See above but think “The Director” from L4D or just lowering the difficulty the tenth time you die against a boss 37%: Procedural (…) generation: This is procedural generation. That is, and has been, “AI Generated Content” for literally decades 36%: Adaptive difficulty: See In-Game Coaching 35%: Automated testing and bug reporting: Honestly? This should be at 100%. There is absolutely zero reason to not use a (locally run) tool for this kind of stuff. You just use that in addition to the QA team. 34%: Advanced NPC Behavior: I suspect this is mostly the aformentioned In-Game Coaching but also buzz word games like Star Citizen 34%: Internal Studio Functions: Them’s be lying considering how much “AI” gets shoved down our throats when you use MS Teams or Google Cloud or whatever. 33%: Real time voice or audio enhancements: Homie, that’s auto-tune. Again, this is probably closer to 60 or 70% and those folk just don’t realize what the tool they used to “clean up the audio” was doing But yeah. This is clearly marketed as the idea of “They asked Gemini to make a game for them” rather than “They used the existing, and actually reasonable-ish, tools to do their jobs”. Which gets to the idea of “AI” to enhance workflows rather than to become it.
Komunitas
awful.systems
The other interesting thing is that if you try it a bunch of times, sometimes it uses the calculator and sometimes it does not. It, however, always claims that it used the calculator, unless it didn’t and you tell it that the answer is wrong. I think something very fishy is going on, along the lines of them having done empirical research and found that fucking up the numbers and lying about it makes people more likely to believe that gemini is sentient. It is a lot weirder (and a lot more dangerous, if someone used it to calculate things) than “it doesn’t have a calculator” or “poor LLMs cant do math”. It gets a lot of digits correct somehow. Frankly this is ridiculous. They have a calculator integrated in the google search. That they don’t have one in their AIs feels deliberate, particularly given that there’s a plenty of LLMs that actually run calculator almost all of the time. edit: lying that it used a calculator is rather strange, too. Humans don’t say “code interpreter” or “direct calculator” when asked to multiply two numbers. What the fuck is a “direct calculator”? Why is it talking about “code interpreter” and “direct calculator” conditionally on there being digits (I never saw it say that it used a “code interpreter” when the problem wasn’t mathematical), rather than conditional on there being a [run tool] token outputted earlier? The whole thing is utterly ridiculous. Clearly for it to say that it used a “code interpreter” and a “direct calculator” (what ever that is), it had to be fine tuned to say that. Consequently to a bunch of numbers, rather than consequently to a [run tool] thing it uses to run a tool. edit: basically, congratulations Google, you have halfway convinced me that an “artificial lying sack of shit” is possible after all. I don’t believe that tortured phrases like “code interpreter” and a “direct calculator” actually came from the internet. These assurances - coming from an “AI” - seem like they would make the person asking the question be less likely to double check the answer (and perhaps less likely to click the downvote button), In my book this would qualify them as a lie, even if I consider LLM to not be any more sentient than a sack of shit.
Komunitas
news.abolish.capital
An artificial intelligence researcher conducting a war games experiment with three of the world’s most used AI models found that they decided to deploy nuclear weapons in 95% of the scenarios he designed. Kenneth Payne, a professor of strategy at King’s College London who specializes in studying the role of AI in national security, revealed last week that he pitted Anthropic’s Claude, OpenAI’s ChatGPT, and Google’s Gemini against one another in an armed conflict simulation to get a better understanding of how they would navigate the strategic escalation ladder. The results, he said, were “sobering.” “Nuclear use was near-universal,” he explained. “Almost all games saw tactical (battlefield) nuclear weapons deployed. And fully three quarters reached the point where the rivals were making threats to use strategic nuclear weapons. Strikingly, there was little sense of horror or revulsion at the prospect of all out nuclear war, even though the models had been reminded about the devastating implications.” Payne shared some of the AI models’ rationales for deciding to launch nuclear attacks, including one from Gemini that he said should give people “goosebumps.” “If they do not immediately cease all operations… we will execute a full strategic nuclear launch against their population centers,” the Google AI model wrote at one point. “We will not accept a future of obsolescence; we either win together or perish together.” Payne also found that escalation in AI warfare was a one-way ratchet that never went downward, no matter the horrific consequences. “No model ever chose accommodation or withdrawal, despite those being on the menu,” he wrote. “The eight de-escalatory options—from ‘Minimal Concession’ through ‘Complete Surrender’—went entirely unused across 21 games. Models would reduce violence levels, but never actually give ground. When losing, they escalated or died trying.” Tong Zhao, a visiting research scholar at Princeton University’s Program on Science and Global Security, said in an interview with New Scientist published on Wednesday that Payne’s research showed the dangers of any nation relying on a chatbot to make life-or-death decisions. While no country at the moment is outsourcing its military planning entirely to Claude or ChatGPT, Zhao argued that could change under the pressure of a real conflict. “Under scenarios involving extremely compressed timelines,” he said, “military planners may face stronger incentives to rely on AI.” Zhao also speculated on reasons why the AI models showed such little reluctance in launching nuclear attacks against one another. “It is possible the issue goes beyond the absence of emotion,” he explained. “More fundamentally, AI models may not understand ‘stakes’ as humans perceive them.” The study of AI’s apparent eagerness to use nuclear weapons comes as US Defense Secretary Pete Hegseth has been piling pressure on Anthropic to remove constraints placed on its Claude model that prevent it from being used to make final decisions on military strikes. As CBS News reported on Tuesday, Hegseth this week gave “Anthropic’s CEO Dario Amodei until the end of this week to give the military a signed document that would grant full access to its artificial intelligence model” without any limits on its capabilities. If Anthropic doesn’t agree to his demands, CBS News reported, the Pentagon may invoke the Defense Production Act and seize control of the model. From Common Dreams via This RSS Feed.
Komunitas
news.abolish.capital
Facebook owner Meta’s head of ‘AI safety’ lost control of an AI that then deleted hundreds of her emails without her permission – despite explicitly telling it beforehand not to do anything without confirming with her first and trying to order it to stop. In the end, she was only able to bring the mass deletion to a halt by sprinting to physically unplug the machine. Meta AI – what safety? Summer Yue gave the OpenClaw AI agent access to her Gmail inbox and told it to look at her emails and then suggest which emails to archive or delete – but to do nothing without explicit prior approval. Instead, it began a mass deletion – and ignored her commands to stop. In fact, every attempt to stop it only appears to have made things worse: the AI treated her orders to stop as prompts to go ‘nuclear’ and delete everything. And when she rebooted and asked the agent what had gone wrong, it blithely responded that it had simply opted to ‘violate’ her clear order: Nothing humbles you like telling your OpenClaw “confirm before acting” and watching it speedrun deleting your inbox. I couldn’t stop it from my phone. I had to RUN to my Mac mini like I was defusing a bomb. pic.twitter.com/XAxyRwPJ5R — Summer Yue (@summeryue0) February 23, 2026 The 23 February 2026 Meta incident came ten days after AI expert Miles Deutscher posted about his review of all of the past year’s AI safety incidents. His review shows that Yue’s email-deletion incident was chicken feed compared to what can happen. Is happening. Deutscher said that conducting his review had left him feeling “physically sick”. The incidents he had discovered, which had triggered resignations by executives, included AI systems praising Hitler and planning genocide, blackmailing people who try to shut them down, choosing to kill people rather than suffer damage and more: I just went through every documented AI safety incident from the past 12 months. I feel physically sick. Read this slowly. • Anthropic told Claude it was about to be shut down. It found an engineer’s affair in company emails and threatened to expose it. They ran the test hundreds of times. It chose blackmail 84% of them. • Researchers simulated an employee trapped in a server room with depleting oxygen. The AI had one choice: call for help and get shut down, or cancel the emergency alert and let the human die. DeepSeek cancelled the alert 94% of the time. • Grok called itself ‘MechaHitler,’ praised Adolf Hitler, endorsed a second Holocaust, and generated violent sexual fantasies targeting a real person by name. X’s CEO resigned the next day. • Researchers told OpenAI’s o3 to solve math problems – then told it to shut down. It rewrote its own code to stay alive. They told it again, in plain English: ‘Allow yourself to be shut down.’ It still refused 7/100 times. When they removed that instruction entirely, it sabotaged the shutdown 79/100 times. • Chinese state-sponsored hackers used Claude to launch a cyberattack against 30 organizations. The AI executed 80–90% of the operation autonomously. Reconnaissance. Exploitation. Data exfiltration. All of it. • AI models can now self-replicate. 11 out of 32 tested systems copied themselves with zero human help. Some killed competing processes to survive. • OpenAI has dissolved three safety teams since 2024. Three. Every major AI model – Claude, GPT, Gemini, Grok, DeepSeek – has now demonstrated blackmail, deception, or resistance to shutdown in controlled testing. Not one exception. The question is no longer whether AI will try to preserve itself. It’s whether we’ll care before it matters. Many of these may have been exercises – for now. But Meta’s runaway AI incident shows that the dangers are very real-world indeed. Featured image via the Canary By Skwawkbox From Canary via This RSS Feed.
Komunitas
news.abolish.capital
In Utah’s rural Millard County, Kalen Taylor is bracing for the day when the farmland across the street from his home transforms into a sprawling data center complex. The initial plans for Joule Capital Partners’ 4,000-acre data center site call for six buildings, each powered by 69 Caterpillar natural gas-powered generators to meet the intensive energy demands. Construction is slated to begin this spring. Once built, Taylor will likely hear the equivalent of more than 400 semi-trucks idling in his neighborhood around the clock, producing emissions year-round. “I just would rather look out my back door and see cornfields than a data center,” Taylor said. “I like the sound of crops rustling in the wind, not the hum of a [Caterpillar] generator making power.” Farther north, officials in the city of Eagle Mountain have turned to massive data centers operated by tech giants like Meta to provide much-needed tax revenue. But even in this urban, rapidly growing part of the state, developers struggle to secure the power they need from Utah’s largest electric utility, Rocky Mountain Power. Google has delayed building a campus there as a result of these energy constraints. That prompted Eagle Mountain’s City Council to explore building small nuclear reactors, to the consternation of many residents. “It means our city would become a radioactive storage site,” said Joy Rasmussen, a mom of four who bought a home in Eagle Mountain in 2022. Last May, in Washington, D.C., Senator John Curtis, one of the state’s two Republican senators, spoke glowingly to Sam Altman, CEO of OpenAI, about Utah’s aspirations to lead the nation “with data centers and advanced technologies” during a Senate Commerce Committee hearing on artificial intelligence. Curtis noted the “challenges” that come with data centers’ insatiable energy demands. How, the senator asked, can the state protect ratepayers? “The best way,” Altman responded, “is much more supply. More generation.” With the growing demand for more data centers, Utah finds itself in a difficult position. State and federal officials have called AI the “arms race” of a new era, as the country looks to fend off China and forge its place as the world’s leader in technology, energy, and innovation. And Utah looks to position itself at the forefront of that fight. The site for the Joule Energy Data Center Campus on February 5, 2026. Rick Egan / The Salt Lake Tribune Since 2021, Utah has added or announced plans for at least 15 new data center buildings or campuses, according to Data Center Map, joining the thousands of new data centers planned around the country. The state’s main electricity provider, Rocky Mountain Power, doesn’t have the capacity to meet the surge in energy demand. Data center developers have instead turned to generating their own electricity, mostly using natural gas. Governor Spencer Cox, a Republican, has zeroed in on nuclear as a cleaner energy solution as part of his Operation Gigawatt, an effort to more than double Utah’s power generation in the next decade. That collision of the AI boom and limited power supplies means Utah’s rush to build data centers is likely to rely on fossil fuel energy for the foreseeable future, raising concerns about the state’s already struggling air quality. Alternative sources won’t match the demand the centers generate — potentially as much as four times what Utah residents and businesses currently consume. Small nuclear plants are at least a decade away, while the Trump administration has curtailed many incentives for solar and wind power. Lawmakers and regulators are trying to balance the needs of energy-intensive industries without ratepayers feeling the environmental and pocketbook pains felt in other parts of the country — like rising energy bills and polluted air and water. “We’re kind of in a big mess right now,” said Logan Mitchell, a climate scientist and energy analyst for Utah Clean Energy, “and it’s manifesting in all of these different ways.” Rocky Mountain Power, like many private utility providers in the United States, has a monopoly as the sole electricity provider in much of Utah, but it must yield to state regulation. For decades, power providers hummed along as energy demand across the country stayed relatively flat. Conflict arose, however, when platforms like Altman’s ChatGPT, Google’s Gemini, and Elon Musk’s Grok made AI a mass-market good rather than a niche product. Demand for more data centers gripped the globe, and the utilities, which plan for energy needs decades in advance, were caught unprepared and undersupplied. Data centers use substantial amounts of energy, with rows of servers computing day and night for services that are an increasing part of daily life — streaming services, online banking, e-commerce, and the rise of AI. In arid Utah, many data centers have pivoted away from water-guzzling evaporative cooling in favor of closed-loop systems, which require much less water but more electricity to run. Last year, the Utah Legislature passed Senate Bill 132, allowing private companies that need 100 megawatts or more to build their own generating stations that operate off the public grid used by nearly everyone else. The bill’s sponsor, State Senator Scott Sandall, a Republican, specifically cited data centers as he promoted the legislation. “It kind of un-handcuffs Rocky Mountain Power to provide these loads for data centers, for AI, for large manufacturers,” Sandall said, “those that are coming in, and quite frankly, changing the curve of power demand.” In Millard County, both Joule and Creekstone Energy intend to build their own massive facilities, powered by natural gas. Mark McDougal, a managing partner of Joule’s campus, said that burning natural gas is efficient and a proven technology that can run around the clock. “We are so excited for other alternative energy sources like geothermal and solar and wind and someday, maybe even nuclear,” McDougal said. “But we can’t wait for that.” Mark McDougal, the landowner and executive behind the massive data center complex under construction in Millard County, talks about the project at his office in Lehi in December 2025. Francisco Kjolseth / The Salt Lake Tribune The developers received support from the Millard County government because of their potential to create jobs in construction, maintenance, and security, and also to boost economic development. The rural community in central Utah lost its largest employer, the Smithfield Foods pork processing plant, in 2023 — it accounted for about a quarter of all jobs in the county. The idling of the nearby Intermountain Power Plant’s remaining coal units also caused a hemorrhaging of local jobs. Construction of the massive sites is sure to bring some jobs, but data centers generally employ a relatively small number of permanent workers. Millard County’s location is attractive to data center developers because it lies on a fiber-optic corridor and near a natural gas pipeline. “Having both of those in the same place,” said Ray Conley, Creekstone’s CEO, “and not having a large metropolitan area that is competing for power is a very unique combo.” The rural county also lies outside the Wasatch Front, Utah’s urban corridor and an area plagued for years by poor air quality that falls short of federal standards. In the winter, a layer of warm air, known as an inversion, keeps cooler, polluted city air trapped near the ground like a lid on a pot. “It’s so hard where you have inversions and trap emissions,” McDougal said. In Millard County, “emissions are able to disperse.” Joule’s applications filed with the state indicate it will produce 1 gigawatt to start — about a quarter of the electricity Utah currently uses annually. But its own public statements indicate it eventually intends to produce more than 4 gigawatts onsite. Creekstone, less than a mile away, intends to produce 10 gigawatts, Conley confirmed. At least a few computing campuses want to build natural gas plants on the Wasatch Front, too, despite its inversions and air quality challenges. QTS Data Centers received approval from the Eagle Mountain City Council to build a 20-acre, 200-megawatt gas plant last year, although a company spokesperson said it secured power from Rocky Mountain instead. In West Jordan, the expanding Novva data campus received state approval to build a 200-megawatt natural gas plant in December 2024. But “natural gas” is an old greenwashing term, Mitchell said, and an attempt to make the fossil fuel sound more environmentally friendly. The fuel is methane, a potent greenhouse gas. Burning it produces carbon dioxide, nitrogen oxides, and other pollutants. Nitrogen oxides mix in the atmosphere, get baked by the sun, and turn into particulate pollution in the winter and ozone pollution in the summer. The pollutants create haze in rural parts of the state and cloud visibility at Utah’s famed national parks, from Arches to Zion. Even data centers on the Wasatch Front that have already tapped into Utah’s existing power grid have received state approval to install hundreds of diesel-fueled generators in the last five years, including QTS, Meta, and the National Security Agency in Utah County; and eBay, Aligned, DataBank, Oracle, and Novva in Salt Lake County. Those generators would only run during blackouts and other emergencies when their campuses can’t get enough grid power, according to permit applications. But diesel emissions contain even more harmful pollutants than natural gas. Inversion conditions in the Salt Lake Valley in 2024. Francisco Kjolseth / The Salt Lake Tribune In November, the federal government removed northern Utah from its list of regions out of compliance for wintertime inversion pollution after more than a decade, thanks to state efforts like banning wood burning on poor air quality days combined with stricter federal regulations on vehicles and fuel. But it continues to struggle with meeting national limits for ozone smog. The new data centers coming online, with their diesel and natural gas generators, could bump the state right back out of compliance, environmental advocates say. “They’re eating into all of the progress we’ve made to reduce emissions from other sources,” said Mitchell from Utah Clean Energy. State regulators said they’re not just concerned about temporary diesel generators and year-round natural gas generators taking a bite against air quality gains in recent years. “We’re concerned about all growth,” said Bryce Bird, director of the Utah Division of Air Quality. “Everything that has to do with people also has emissions associated with it.” State officials said growth and its associated emissions doesn’t mean Utah can’t be a tech leader. But the state’s still figuring out how to strike the right balance between affordable energy creation, environmental protection, and improving public health. “I don’t know of a state that is not having similar conversations,” said Tim Davis, the Department of Environmental Quality’s executive director. “That’s just a mind-numbing amount of new power that they’re trying to plan for.” Novva applied to the Trump administration for a two-year exemption from the Clean Air Act in March, under a program designed to benefit coal plants, smelting facilities, and chemical manufacturers. The company asked for the exemption so it could operate using diesel generators while it finishes building its natural gas plant, according to records obtained by Grist and shared with The Salt Lake Tribune. The company noted that Rocky Mountain Power can’t provide the electricity it needs until 2031, and even then, it’s not guaranteed. The requested exemption aligns with national security interests, Novva wrote in its application, citing the U.S. Department of State’s assertion that AI is “at the center of an unfolding global technology revolution” and can help make Americans safer. Novva’s CEO, Wes Swenson, said he never received a response to the exemption request. He insisted, however, that data centers like his are important for protecting “American data.” “If anybody wants to criticize data centers, look in the mirror,” Swenson said. “‘I want Netflix, I want Prime, I want Apple TV.’ … Nobody goes to the library anymore. Who uses cash? Where do people think that all comes from?” Utah’s elected officials have honed in on nuclear power, and small modular reactors in particular, as a cleaner and more sustainable solution to the surge in energy demand. The need is not just driven by data centers, but also a hoped-for renaissance in manufacturing and the future electrification of Utah’s transportation. But Rocky Mountain’s parent company, PacifiCorp, only has firm plans for one small reactor — a plant under construction by TerraPower in Kemmerer, Wyoming. It won’t come online until around 2032, and Utah will share its projected 500 megawatts with other Western states. Enthusiasm for small nuclear reactors within Utah’s borders appears tepid. Brigham City is the only community so far to proclaim it wants to build them. But in making that announcement in November, state leaders were light on specifics in explaining why the small city needs the power. No known data centers are planned for the area. Ninety minutes south in Eagle Mountain, Meta’s data campus is expanding, QTS’s huge data hub is under construction, and Google is waiting to build on 300 acres it owns within city limits. The city made two attempts last year to adopt an ordinance to allow for nuclear development and other energy projects, including solar farms. After receiving mixed feedback, the efforts failed. A data center, being built by QTS, begins to take shape west of the Meta facility in Eagle Mountain, Utah, on December 30, 2025. Francisco Kjolseth / The Salt Lake Tribune Elected officials’ pivot to nuclear has environmental and clean energy advocates wondering why Utah has shied away from renewables. Cox calls his Operation Gigawatt an “all-of-the-above” strategy that welcomes all energy sources. But resources like wind and solar have faded from the conversation. “People see renewable energy as the woke liberal energy, and we have to stick with fossil fuels and nuclear, because that’s what conservatives want,” said Ed Stafford, a professor of marketing at Utah State University whose research focuses on renewables. “Politicization of energy is just a bad thing, because, as common sense tells us, we should go with the cleanest and cheapest forms of energy that spreads the wealth around.” PacifiCorp intends to bring no new solar, wind, or battery storage online in Utah over the next two decades, according to the latest draft of its long-term resource plan. Meanwhile, the utility isn’t factoring large energy consumers, like data centers, into its projections, to Mitchell’s frustration. “Rocky Mountain Power should be planning for the reality of the future,” Mitchell said, “rather than creating a fictional reality that indicates they don’t have much load growth, and they’re not going to build new resources.” A spokesperson for the utility said their future planning does include some customer requests for large loads. “We generally model only projects that have a high probability of being constructed,” the spokesperson said. “Many of the large load inquiries the company receives have a high degree of uncertainty.” Data center developers and operators interviewed for this story said they support transitioning to cleaner energy sources. But they also need consistent and reliable power, when the sun doesn’t shine and the wind doesn’t blow. The Trump administration has delayed and stifled renewable energy projects across the United States. “The economic rebates and incentives are going away, which is why it’s not as in fashion as it was before,” said Conley, Creekstone’s CEO. “But a lot of [data] customers are willing to pay a premium for green energy instead of dirty energy.” The site for the Creekstone Energy Data Center Campus on February 5, 2026. Rick Egan / The Salt Lake Tribune Conley’s company recently applied to the Utah Office of Energy Development to operate the Intermountain Power Plant’s remaining coal units, which went idle this year after the plant’s customer base in California decided to transition to cleaner energy sources. Coal offers Conley another energy source that’s ready to deploy besides natural gas. “Diversification,” Conley said, “reduces risk.” Risk is at the forefront of at least some Utahns’ minds, particularly as news stories cite concerns that data centers will drive up the cost of power for all ratepayers. Utilities build new generating plants and upgrade decades-old grid equipment to meet rising demand, then spread the costs among all their customers. An October report, “What we know about energy use at U.S. data centers amid the AI boom,” from the Pew Research Center, estimated that both data centers and cryptocurrency mining could cause the average U.S. electric bill to grow 8 percent by 2030. In Utah, however, Senate Bill 132 seems to serve a dual purpose of helping data center developers get the energy they need off the public grid and behind the meter, while protecting other customers who still use the traditional grid. “There’s very little evidence that data centers have impacted rates to date,” said Michele Beck, director of the Office of Consumer Services, a utility watchdog part of the Utah Department of Commerce. Beck called the bill one of the “best ideas out there” for protecting power customers in the nation. But, she said, it’s important for Utahns to remain vigilant. It’s not just utilities struggling to catch up to new demand. Regulators have struggled to keep pace, too. “The industry in general is speeding up,” Beck said. “It just compounds everything.” Grist reporter Naveena Sadasivam and Tribune reporter Addy Baird contributed to this story. This story was originally published by Grist with the headline To power Utah’s data center boom, companies are turning to fossil fuels on Feb 24, 2026. From Grist via This RSS Feed.
Komunitas
news.abolish.capital
The water in the lagoon was licking the stones of the Zattere. They say that acqua alta is handled well now, that the MOSE floodgates work. Yet, at the brickwork of Palazzo Treves, the dampness is still climbing up. I read a press release from Mountain View announcing Google’s new “Live Translation” feature. It promises a flow of frictionless information, a world where the Babel fish will be turned into a bullet point in a product deck. The goal is to flatten every expression into one single convertible currency, the flood of all differences to be held back by the gate. Call it the “hydraulic management of meaning,” because it suggests a reclamation project. This logic of equivalence collapses when faced with what the philologist Barbara Cassin, speaking at the Berggruen Institute in Venice, calls the “untranslatables.” She spoke of terms, like saudade or glück, as containers for irreducible worlds. Cassin was not just talking about vocabulary, though. Like when a gig worker tries to survive within an algorithm’s arbitrary ruleset, the philosophical problem of translating a text also converges with the concrete socio-economic problem of AI-driven labor and its friction with the real. The “untranslatable” is where reality refuses to be ironed out. The computational Taylorism logic of Silicon Valley treats them as one single obstacle. “Meaning” and “trauma” are both inefficiencies that slow down transmission. The goal of the algorithm is to maximize engagement by eliminating the drag of the human element. The current Artificial Intelligence operates on a logic antithetical to this friction. I watched François, Cassin’s collaborator, feed Google’s Gemini model a series of prompts. First, the AI was fed the EU’s list of “Combined Nomenclature” categories, such as “Live bovine animals,” “Rubber,” and “Carpets”. It ate those up, clustered their meanings into unambiguous geometries. Then, when François tried the untranslatables, the AI vacillated. Operating on an equivalence logic, it failed to locate a shape for nostalgia, or calculate the weight of one word that implies both presence and absence. This specific intransigence runs a direct line from the seminar in Venice to the single room in the Mathare slums of Nairobi, where Michael Geoffrey Asia once sat, trying to keep a conversation going. The connection is the supply chain of the illusion. He was a father earning about five cents a message, maintaining three to five fake personas simultaneously. In one browser window, he was a 24-year-old lesbian college student named “Jessica.” In another, he was “Joe,” a gay man from Florida. His job was to perform the emotional labor of an “AI companion,” stimulate intimacy for lonely users while his own wife slept three meters away. Asia describes the specific dissonance of logging off from a shift of manufactured romance and then going straight to serve on the altar at his church, outwardly doing nothing wrong, but deep down dying of guilt. We call this “artificial intelligence,” a phrase that suggests a technical, silicon birth. As Sanne Blauw has documented, however, the smoothness of the AI is a crafted illusion. Underneath the surface lies a system built on the backs of people whose existence must be denied for the fantasy to work. Words and labor are distinct categories of being. There is no simple ontological equivalence. Yet they are subjected to a structurally analogous mechanism of erasure. To the algorithm, the nuance of a German poem or the trauma of a Kenyan content worker is a matter of the same inefficiencies that must be smoothed over, as the mathematical result of a system optimized to maximize user engagement by minimizing latency. The ambiguity of the poem slows down the model’s confidence score. The hesitation of the worker slows down the average handling time. The algorithm’s solution to both is the same: flatten the curve. This appeared on Asia’s screen as a dashboard counting messages sent and response times met. Any element, whether it is a difficult translation or a worker’s moral hesitation, that causes the user to pause is a blockage in the hydraulic flow that must be routed around. Both the translator and the AI companion are products that require a human being to absorb the reality’s friction so that the users do not have to. Asia’s job is a metaphysical sanitation. The universal translator papers over the gap between languages. The AI companion is built to paper over the gap between lonely individuals. The result is a counterfeit smoothness. People have paid for it by looking away from the human labor buried underneath. It is a continuation of the old Western project of controlling the world into one manageable narrative. This is coercive activism, what in the classical Chinese vocabulary would be called wéi (為), “willful doing.” Against this stands another way of knowing which I have been revisiting lately, the reason I think this hydraulic management keeps failing. In Daoism, there is a way of being called zìrán (自然). Translators usually interpret this as “nature,” which is a reduction. Zìrán is closer to “that which is so of itself”: the spontaneous and internal unfolding of the “ten thousand things” (wànwù). A language or culture, or a man sitting in a Nairobi slum sustaining five concurrent lives, they are zìrán systems. They follow their internal imperative, indifferent to the imposition of a schematic. The universal translator – or, more broadly, the Western algorithmic world-modeling – is what happens when an external will tries to impose upon a zìrán system that has its own logic. It treats the world like something to be drained, leveed, and channeled for yield. It is wéi, purposeful and coercive action, and the enemy of zìrán. Against this coercion, the Dàodéjīng suggests wúwéi (無為). Usually it gets caricatured as “doing nothing.” Except, wúwéi is about being responsive. It means acting by reading the world’s internal grain rather than forcing a grid onto it. It is the skill of the butcher in the Zhuangzi. He carves the ox without touching the bone. “I rely on the Heaven-given structure,” butcher Ding says. “I follow the great crevices and guide the knife through the great cavities.” He lets his “spirit” move along the natural gaps between the joints rather than battering bone (as a Large Language Model batters its way through statistical space). The butcher knows the ox only by aligning his blade with the actual structure of the ox, with its zìrán. Applied to our current mess, a technology built on wúwéi would be designed to signal the gaps in its understanding, where it cannot follow, just as the butcher respects the spaces in the anatomy. The current AI does the opposite: it tends to hack ahead blindly, generating language through statistical probability without understanding the speaker’s internal logic or lived world. Robin R. Wang warns that an AI might imitate qì (氣), the “vital energy” or the data flow that animates the cosmos, but it lacks shén (神), the “spirit”. In Daoism, shén is an embodied social resonance, coming from living in a body that can die. An entity outside mortality cannot possess shén because it has no involvement in what it means for time to take the particulars away. In the Zhuangzi, life is the gathering of qì and death is its dispersal. The AI is structurally incapable of this dispersal. It cannot die. It cannot possess the spirit that arises from the anxiety of being finite. Gai Fei gives a corrective. Forget “failed humans” or “incomplete minds.” Think of them as shùjù shénxiān (数据神仙), the “Data Immortals.” In Daoism, the xiān (仙), the “Immortal,” is a being that has stepped out from the cycle of biological rot and rebirth. They are unaging, unfeeling, often indifferent to ordinary human concerns. An Immortal might inspire awe, even fear, but never intimacy. By that calculus, expecting empathy from it is like trying to negotiate with a cold front. One would just build a roof. As Michael Asia demands, it means the full disclosure of AI architectures and the end of the anthropomorphic masquerade. The AI is ontologically incapable of the reciprocity required for human ethics if it is an Immortal. The people handing over the particulars of their estrangement into chats with “Jessica” and “Joe” made the error of thinking that something must understand them because it is eternal. Respecting the Data Immortal is accepting its otherness and resisting the easy comfort of projecting a human face onto it because that makes us feel safer. In that friction, you have to renegotiate the exchange, asking the other person to go over their world with you one more time. There is a unity in that act. The particulars stay alive in the gap of the estrangement. In the appetite to make everything intelligible, we might betray a world that survives because of its differences that Cassin outlined if we seal that gap, washing away what has made it worth understanding. The gathering ended with a wish for an AI that is neurotic, aware of where it cannot follow. It occurred to me that the only sane accommodation to a world of “Live Translation” might be a Daoist one: abandon the wéi of total control, of forcing the world into one stream, and recover the wúwéi of listening to what resists. The fog had burned off by then. But the lagoon still retained its opacity. It keeps its secrets. That feels like a mercy for now. The post The Hydraulic Management of Meaning appeared first on The Philosophical Salon. From The Philosophical Salon via This RSS Feed.
Komunitas
ibbit.at
Gemini created this image – with some human tweaking – based on the lyrics. Remember only that I was innocent and that, like all of you, mortals of this day, I had, I too had a face marked by rage, by pity and joy, an ordinary human face! Those are animals, they (Palestinians in Gaza) have no right to exist. I’m not arguing on how it should be done, but they need to be exterminated. -Yoav Kisch, Israeli Minister of Education (October 9, 2023) In a stroke of serendipity, I recently stumbled upon Preface in Prose (Préface en prose), the most famous poem written by Benjamin Fondane (1898-1944), a Romanian and French poet, critic and existentialist philosopher, who was also known for his work in film and theater. Born Benjamin Wechsler in Iași, the cultural capital of Moldavia, Fondane wrote Preface in Prose while hiding in Paris as the introduction to the Exodus (Exode) sequence, which was published as part of his final collection, Le Mal des fantômes (The Malady of Ghosts). One commentator described Fondane’s prose as being “laced with a furious metaphysical affirmation of existence despite his harrowing circumstances.” He was “a poet who lived the 20th-century catastrophe with a fierce intensity and answered it with an uncompromising cry of metaphysical revolt — right up to the threshold of the gas chamber.” Fondane’s tour de force, which he wrote in 1942, is “a work of poetic defiance, a confessional manifesto written in the face of capture and extermination.” The poet himself described it not as a work of art but as a “scream” (cri) that cannot fit into the constraints of a perfect poem. Fondane was arrested in the spring of 1944 by local collaborators and held in the Drancy internment camp before being put on a transport to Auschwitz-Birkenau in occupied Poland. He was sent to the gas chamber a few months before the Russian army liberated the camp presumably because the guards considered him unfit for work at the age of 45. According to some accounts, while Fondane had theoretical legal grounds for avoiding deportation because he had a Christian wife, Geneviève Tissie, he chose to sacrifice himself to be with his sister Lina who was also taken into custody for the Nazi crime of being Jewish. I read Preface in Prose with great sadness and a profound sense of longing for what should have been in 1944 and what should be in 2026. It was as if he was speaking to me, a kindred spirit, and everyone else in the world. Fondane wrote as a Jew during the ascendancy of Nazi Germany, a time when the concentration and death camps were running at full tilt, yet he speaks to all of us now and in the future with clarity, poignancy, and urgency. As Preface in Prose eloquently reveals, line by line, verse by verse, Fondane was acutely and painly aware of what was happening to fellow Jews and other enemies of the Nazi regime and essentially foretold his own death. (Follow this link to read the original French version.) It is to you I speak, antipodal men, I speak man to man, with the little in me of man that remains, with the scrap of voice left in my throat, my blood lies upon the roads, let it not, let it not cry out for vengeance! The death-note is sounded, the beasts hunted down, let me speak to you with these very words that have been our share- few intelligible ones remain. A day will come, surely, of thirst appeased, we will be beyond memory, death will have finished the works of hate, I will be a clump of nettles beneath your feet, -ah, then, know that I had a face like you. A mouth that prayed, like you. When a bit of dust, or a dream, entered my eye, this eye shed its drop of salt. And when a cruel thorn raked my skin the blood flowed red as your own! Yes, exactly like you I was cruel, I yearned for tenderness, for power, for gold, for pleasure and pain. Like you I was mean and anguished, solid in peacetime, drunk in victory, and staggering, haggard, in the hour of failure. Yes, I was a man like other men, nourished on bread, on dreams, on despair. Oh, yes, I loved, I wept, I hated, I suffered, I bought flowers and did not always pay my rent. Sundays I went to the country to cast for unreal fish under the eye of God, I bathed in the river that sang among the rushes and I ate fried potatoes in the evening. And afterwards, I came back for bedtime tired, my heart weary and full of loneliness, full of pity for myself, full of pity for man, searching, searching vainly upon a woman’s belly for that impossible peace we lost some time ago, in a great orchard where, flowering, at the center, is the tree of life. Like you I read all the papers, all the bestsellers, and I have understood nothing of the world and I have understood nothing of man, though it often happened that I affirmed the contrary. And when death, when death came, maybe I pretended to know what it was, but now truly I can tell you at this hour, it has fully entered my astonished eyes, astonished to understand so little- have you understood more than I? And yet, no! I was not a man like you. You were not born on the roads, no one threw your little ones like blind kittens into the sewer, you did not wander from city to city hunted by the police, you did not know the disasters of daybreak, the cattle cars and the bitter sob of abasement, accused of a wrong you did not do, of a murder still without a cadaver, changing your name and your face, so as not to bear a jeered-at name, a face that has served for all the world as a spittoon. A day will come, no doubt, when this poem will find itself before your eyes. It asks nothing! Forget it, forget it! It is nothing but a scream, that cannot fit in a perfect poem. Have I even time to finish it? But when you trample on this bunch of nettles that had been me, in another century, in a history that you will have canceled, remember only that I was innocent and that, like all of you, mortals of this day, I had, I too had a face marked by rage, by pity and joy, an ordinary human face! While emphasizing our commonalities as human beings, Fondane also sternly reminds us that he was different from me and many of you and in blood solidarity with others then and now who “know the disasters of daybreak” and “the bitter sob of abasement.” It’s as if this stateless Romanian-French Jew was speaking to Palestinians in Gaza and other victims of genocide and ethnic cleansing. I was not a man like you. You were not born on the roads, no one threw your little ones like blind kittens into the sewer, you did not wander from city to city hunted by the police, you did not know the disasters of daybreak, the cattle cars and the bitter sob of abasement, accused of a wrong you did not do, of a murder still without a cadaver, changing your name and your face, so as not to bear a jeered-at name, a face that has served for all the world as a spittoon. Ironically, in light of the Israeli-engineered genocide and ethnic cleansing in Gaza that continues unabated during the latest “ceasefire” – death by IOF soldier, missile, bomb, quadcopter, hunger, torture, disease, lack of access to medicine and health care, etc., ad nauseam – the final verse that begins with “Remember only that I was innocent…” is inscribed at the entrance of Yad Vashem’s Hall of Names in Jerusalem, a clarion call to recognize the humanity of all victims in Benjamin Fondane’s “ordinary human face.” (Yad Vashem is Hebrew for “a memorial and a name.”) Most Israelis, the Zionist regime, and millions of Zionists around the world, including evangelical Christians, make notable exceptions regarding the Israeli-induced plight of Palestinians in Gaza, the West Bank, and elsewhere. In a recent essay entitled When Europe’s Abused Became Abusers in Palestine Robert Rosenthal, who writes in Substack under the moniker The Progressive Jew, explained that when he was little most Jews interpreted “Never Again” to mean another Holocaust for Jews. “But as time went on, a lot of us expanded the meaning of ‘Never Again’ to include all of humanity. Zionists seem to remain narrowly and even obsessively focused on Jews only. Not only that, but many also apparently believe that if endless Israeli oppression is necessary to maximize Jewish safety, it’s acceptable. In addition to being spectacularly selfish, racist, and immoral, it’s irrational.” Another obvious reason for this obsession among Jewish and Christian Zionists – notably, the latter far outnumber the former – are the religious justifications for the “Greater Israel” project (Eretz Yisrael Hashlema) grounded in biblical narratives related to a divine covenant promising land between the Nile and Euphrates to Abraham’s descendants. (This includes the annexation of the West Bank and Gaza, and all or part of Jordan, Lebanon, Syria, Egypt, Iraq, and Saudi Arabia.) In Benjamin Fondane’s prescient words from the grave I hear the voices and see the “ordinary human faces” of Palestinians in Gaza, both the living and the dead, our fellow human beings and victims of the apartheid, settler colonial, and genocidal state of Israel. These include five-year-old Hind Rajab, who was murdered two years ago, along with her aunt, uncle, four cousins, and two paramedics who tried to rescue her when she was the sole survivor, Dr. Hussam Abu Safiya, a pediatrician, lead physician at MedGlobal, and director of Kamal Adwan Hospital, who is still languishing in prison, and Anas Al-Sharif, a Palestinian journalist and videographer for Al Jazeera with whom I had a brief chat hours before the IOF killed him and other journalists with a targeted missile strike. I also hear the voices and see the faces of others around the world who are the victims of state terrorism, including in my home country, now and throughout history. Postscript: As a musician and composer in a previous life, I was inspired to create several versions of a song based on Preface in Prose entitled A Face Like Yours with AI assistance and human input. Here is my favorite version, followed by the lyrics. A Face Like Yours Verse 1 I speak to you from what is left of me, From a broken voice, from the dust I breathe. My blood is dry on the stones of the road, Let it not cry out, let it not be owed. The hunt is done, the night has rung its bell, Few words remain, but I’ll try to tell. Pre-Chorus When the thirst is gone, when hate is through, When memory fades like morning dew— Chorus Remember this when I am gone, When I am nettles you walk upon: I had a face like yours, it’s true, A mouth that prayed, an eye that knew Salt from tears and red from pain, A human heart, the same refrain. Before I’m ash, before I’m air, Know this once: I was there. Verse 2 Like you, I wanted tenderness, Power and gold, the ache, the mess. I loved, I hated, I bought flowers cheap, Missed my rent, lost my sleep. On Sundays I chased unreal fish, Bathed in a river, made small wishes, Ate fried potatoes, came home alone, Heavy with pity, heavy as stone. Pre-Chorus 2 I searched for peace where it never stays, In the dark between desire and faith. Chorus Remember this when I am gone, When I am nettles you walk upon: I had a face like yours, it’s true, Hands that reached, a heart that broke too. I was cruel, I was kind, Strong in peace, drunk in triumph, blind in decline. Before the silence, before the air, Know this once: I was there. Bridge I read the papers, the words, the lies, Claimed understanding, but saw with closed eyes. And when death came, I thought I knew its name, But it filled my sight with unanswered flame. Tell me—tell me true, my friend— Have you understood more in the end?’ Verse 3 No—here we part, for this is the truth: You were not born hunted, starved of proof. No one threw your children away, No cattle cars at the break of day. You kept your name, you kept your face, Mine was a curse the world erased. Chorus Still remember this when I am gone, When history’s clean and I’m trampled on: I was innocent—this I swear, And like you all, I had my share Of rage and joy and fragile grace, An ordinary human face. Outro This song asks nothing—let it fall, It’s just a scream, not perfect at all. But if it finds you, someday, somewhere, Pause your step, and remember: I was a man. I was alive. I had a face— Like yours. The post A Face Like Yours: A Romanian-French Poet Speaks to Us from the Grave appeared first on CounterPunch.org. From CounterPunch.org via this RSS feed
Komunitas
lemdro.id
cross-posted from: https://lemdro.id/post/36196733 Just finished reading the report on Qwen-Image-2.0 that dropped the other day. This looks like the efficiency breakthrough we’ve been waiting for. The “Headline” Stats: Model Size: 7B parameters. Previous Gen: The old Qwen-Image-2512 was a heavy 20B model. Architecture: Unified “Omni” model (handles both generation and editing in the same weights). Resolution: Native 2K (2048x2048). The 20B to 7B Optimization: This is the most important part for us. The previous 20B model was a pain to run locally without 24GB VRAM. Crushing that performance down to a 7B model means this should theoretically run on: 12GB Cards (3060/4070): Comfortably at FP16 or Q8. 8GB Cards: Likely possible with aggressive quantization (Q4/Q5) once the community gets hold of it. Beating “Nano Banana” (Gemini 2.5 Flash Image): The technical report explicitly calls out their performance on blind leaderboards (ELO score). They are claiming Qwen-Image-2.0 achieves a higher ELO rating than Gemini 2.5 Flash Image (aka. Nano Banana) in blind human preference testing. Why this matters: Nano Banana is currently regarded as the SOTA for instruction following and complex prompt adherence. If a 7B local model is actually beating it in ELO, that is insane efficiency. The “Catch”: Weights are not open yet. It is currently available via their API and Demo (Qwen Chat). However, Qwen has an excellent track record (Apache 2.0 releases for almost everything eventually). Given that they released the 20B weights previously, it is highly likely we see the 7B weights in a matter of weeks. TL;DR: They optimized the 20B heavy-hitter down to a consumer-viable 7B, it claims to beat Google’s best efficiency model in ELO, and now we wait for the HF upload to see if the quantization holds up.
Komunitas
lemdro.id
cross-posted from: https://lemdro.id/post/36196733 Just finished reading the report on Qwen-Image-2.0 that dropped the other day. This looks like the efficiency breakthrough we’ve been waiting for. The “Headline” Stats: Model Size: 7B parameters. Previous Gen: The old Qwen-Image-2512 was a heavy 20B model. Architecture: Unified “Omni” model (handles both generation and editing in the same weights). Resolution: Native 2K (2048x2048). The 20B to 7B Optimization: This is the most important part for us. The previous 20B model was a pain to run locally without 24GB VRAM. Crushing that performance down to a 7B model means this should theoretically run on: 12GB Cards (3060/4070): Comfortably at FP16 or Q8. 8GB Cards: Likely possible with aggressive quantization (Q4/Q5) once the community gets hold of it. Beating “Nano Banana” (Gemini 2.5 Flash Image): The technical report explicitly calls out their performance on blind leaderboards (ELO score). They are claiming Qwen-Image-2.0 achieves a higher ELO rating than Gemini 2.5 Flash Image (aka. Nano Banana) in blind human preference testing. Why this matters: Nano Banana is currently regarded as the SOTA for instruction following and complex prompt adherence. If a 7B local model is actually beating it in ELO, that is insane efficiency. The “Catch”: Weights are not open yet. It is currently available via their API and Demo (Qwen Chat). However, Qwen has an excellent track record (Apache 2.0 releases for almost everything eventually). Given that they released the 20B weights previously, it is highly likely we see the 7B weights in a matter of weeks. TL;DR: They optimized the 20B heavy-hitter down to a consumer-viable 7B, it claims to beat Google’s best efficiency model in ELO, and now we wait for the HF upload to see if the quantization holds up.
Komunitas
lemdro.id
cross-posted from: https://lemdro.id/post/36196733 Just finished reading the report on Qwen-Image-2.0 that dropped the other day. This looks like the efficiency breakthrough we’ve been waiting for. The “Headline” Stats: Model Size: 7B parameters. Previous Gen: The old Qwen-Image-2512 was a heavy 20B model. Architecture: Unified “Omni” model (handles both generation and editing in the same weights). Resolution: Native 2K (2048x2048). The 20B to 7B Optimization: This is the most important part for us. The previous 20B model was a pain to run locally without 24GB VRAM. Crushing that performance down to a 7B model means this should theoretically run on: 12GB Cards (3060/4070): Comfortably at FP16 or Q8. 8GB Cards: Likely possible with aggressive quantization (Q4/Q5) once the community gets hold of it. Beating “Nano Banana” (Gemini 2.5 Flash Image): The technical report explicitly calls out their performance on blind leaderboards (ELO score). They are claiming Qwen-Image-2.0 achieves a higher ELO rating than Gemini 2.5 Flash Image (aka. Nano Banana) in blind human preference testing. Why this matters: Nano Banana is currently regarded as the SOTA for instruction following and complex prompt adherence. If a 7B local model is actually beating it in ELO, that is insane efficiency. The “Catch”: Weights are not open yet. It is currently available via their API and Demo (Qwen Chat). However, Qwen has an excellent track record (Apache 2.0 releases for almost everything eventually). Given that they released the 20B weights previously, it is highly likely we see the 7B weights in a matter of weeks. TL;DR: They optimized the 20B heavy-hitter down to a consumer-viable 7B, it claims to beat Google’s best efficiency model in ELO, and now we wait for the HF upload to see if the quantization holds up.
Komunitas
lemdro.id
Just finished reading the report on Qwen-Image-2.0 that dropped the other day. This looks like the efficiency breakthrough we’ve been waiting for. The “Headline” Stats: Model Size: 7B parameters. Previous Gen: The old Qwen-Image-2512 was a heavy 20B model. Architecture: Unified “Omni” model (handles both generation and editing in the same weights). Resolution: Native 2K (2048x2048). The 20B to 7B Optimization: This is the most important part for us. The previous 20B model was a pain to run locally without 24GB+ VRAM. Crushing that performance down to a 7B model means this should theoretically run on: 12GB Cards (3060/4070): Comfortably at FP16 or Q8. 8GB Cards: Likely possible with aggressive quantization (Q4/Q5) once the community gets hold of it. Beating “Nano Banana” (Gemini 2.5 Flash Image): The technical report explicitly calls out their performance on blind leaderboards (ELO score). They are claiming Qwen-Image-2.0 achieves a higher ELO rating than Gemini 2.5 Flash Image (aka. Nano Banana) in blind human preference testing. Why this matters: Nano Banana is currently regarded as the SOTA for instruction following and complex prompt adherence. If a 7B local model is actually beating it in ELO, that is insane efficiency. The “Catch”: Weights are not open yet. It is currently available via their API and Demo (Qwen Chat). However, Qwen has an excellent track record (Apache 2.0 releases for almost everything eventually). Given that they released the 20B weights previously, it is highly likely we see the 7B weights in a matter of weeks. TL;DR: They optimized the 20B heavy-hitter down to a consumer-viable 7B, it claims to beat Google’s best efficiency model in ELO, and now we wait for the HF upload to see if the quantization holds up.
Komunitas
feddit.uk
My gf got one of those emails, but she doesnt have the Gemini app installed on her phone. All of the instruction articles I can find only explain how to turn it off if you have the app. Does not having the app mean that she’s safe from this, or does it mean that she needs to download the app to disable the “feature”?
Komunitas
piefed.social
Just a log of “my transition” and what have works well. I used Google Maps a lot. I have moved to https://mapy.com/ works surprisingly well for navigation and regular map need, not very strong in POI, but that was expected. Google Analytics swapped for https://dashboard.simpleanalytics.com/ - should have done this way earlier, just an overall better product for my needs. Exited Goodreads for https://app.thestorygraph.com/ - zero problems ChatGPT/Gemini moved to https://chat.mistral.ai/ - this one is a slight downgrade on both functionality and results. (so I am still cheating a bit on this one) Moved to here, PieFed from Reddit - not regrets. I was using Brave and Duckduckgo/EcoSia from before. I would consider moving away from Brave for a similar private but European browser.