Wildroot

Queries may use an external AI service. Details

Wildroot Browser

TopicsAIPage 2

AI · page 2

Further down today’s ranking of AI: page 2 of 7.

How it worksBuild a feed from AI

  1. @dawid@vebinet.com

    The Cambridge Dictionary has just named "human" as its Word of the Year 2026. Of all the words that could have captured this moment, with AI advancing faster than any technology in history, lexicographers chose to remind us of the one thing we should never lose sight of: ourselves. I find this deeply encouraging. As frontier technologies reshape how we work, communicate, and make decisions, we must never forget a fundamental principle: the role of technology is to be human-centered. The true measure of progress is not how sophisticated our tools become, but how much they improve human lives.

    Show more of this post

    Artificial intelligence should be built as assistive technology, a technology that enhances lives, helps solve real problems, and benefits humanity as a whole. AI is not a replacement for a human being, and it should never be treated as one. Judgment, empathy, creativity, accountability: these remain irreplaceably humans. There is also a broader responsibility. Every technological development should serve the purpose of delivering sustainable social and economic development across the world. Empowering people and societies, reducing inequalities and inequities, and ensuring that progress is measured not only in efficiency gains, but in the dignity and opportunity it creates. Technology is at its best when it amplifies human potential, not when it competes with it. We have entered into an AI-driven era, but we should ensure that "human" remains not just the Word of the Year, but the standard by which the technology is being built on. dictionaryblog.cambridge.org/2… #society #AI #technology #ArtificialIntelligence #tech #WordOfTheYear

    Opinion
  2. @9x0rg@mamot.fr

    The research is the latest indictment exposing the lax privacy and security that’s endemic in much of the AI ecosystem. The clearest example involves a Chinese AI lab. This lab’s systems fetched the secret URLs from more than half of the prompts that we sent to its models, including roughly two-thirds of the requests marked as “no data collection” or Zero Data Retention. **Researcher plants 2m canaries into the AI supply chain. A privacy disaster surfaced.** arstechnica.com/security/2026/…

  3. ASA: hey, we signed this Charter of Ethical AI Use in the Creative Industries ME: But there is no ethical use of AI in the creative industries! ASA: We know, but what can you do? The UK SoA withdrew from the charter after a backlash from members. Aussie authors, where are you? #auslit #booksky

    An email to the Australian Society of Authors Hi Saranya Thank you for your reply. You state that you agree there is inherently no ethical use of AI, however the charter explicitly states that ethical AI use is possible. It's in the title! I understand you're trying to protect authors who are already impacted by the illegal use of AI, and I appreciate the work you're doing with the "Not AI's to Take" campaign and other advocate work you do. Your advocacy is one of the main reasons I'm a member. However by signing this charter you are ceding an enormous amount of ground to big tech and to publishers. A voluntary code of practice with no measurable goal or required actions, relies on the majority of people involved being ethical in the first place. Since GenAI is fundamentally unethical (as you've agreed) it's naive in the extreme to expect its adopters to follow this code. I firmly believe you are taking the wrong stance and I tender my resignation, effective immediately. Re
    Opinion
  4. Whenever I read a mathematics textbook, I run into the same problem: the author leaves out details, apparently assuming I should already know them. (Then again, why the hell did I buy the book in the first place?) Or perhaps those missing details were revealed exclusively to his inner circle of PhD students. Well, with AI around, even they are an endangered species now. 😂 Anyway, after all these years of pursuing mathematics as a hobby, I've come to one conclusion: a university-level mathematics textbook is no walk in the park. I want to wish good luck to everyone trying to read the 17 million lines of Lean code Claude just spat out. After all these years, mathematics has finally found its nemesis: an AI that doesn't get to skip the damn steps. And oh, what a glorious day! 😂 #RiemannHypothesis

    Opinion
  5. @markhughes@hachyderm.io

    What happens in AI chat doesn't stay in AI chat. Oops, all your secrets are no longer secrets, a researcher found: - Evidence that at least three AI providers were compromised by attackers who used their access to siphon prompts they received - Multiple resellers in the inference marketplace that logged the prompts and somehow passed them along to people who used the credentials - Multiple cases of AI providers not following end-user privacy settings embedded into flags specifying terms such as Zero Data Retention (ZDR) or “no data collection” arstechnica.com/security/2026/… #LLM #Privacy #Security #AI

    Info
  6. @josephlchu.bsky.social

    Some comments about the whole "AI solved the Riemann Hypothesis" thing... First, notably, unlike the 10,000 AI agent swarm that OpenAI used on Navier Stokes, this one was a famous mathematician employed by Anthropic working with Claude in a kind of human-AI collaboration.

    Opinion
  7. @pschwahn@mathstodon.xyz

    Yes, I get it, we are all crashing out over the supposed AI proof of the Riemann hypothesis. Does the Lean code posted by Alpöge really verify that RH is true? I bet it does! I think the probability of it sneakily exploiting a bug is really low. Do I consider RH solved? Not really - since we don't have a solution! The Lean code has not been read, let alone understood, by a human. But that doesn't mean the proof certificate is useless: at least people know that they can stop trying to disprove RH. There is the almost finished proof by Masullo, which is obviously far more valuable than the AI Lean proof.

    Show more of this post

    Reading Alpöge's remark, he seems pretty genuine and transparent about this. Sure, he can be criticized because he works for an evil tech company (I mean that without sarcasm). But he expressly didn't release a thousands-of-pages slop PDF, so why are people acting like he did, or like his contribution is worthless or didn't require effort? I trust the mathematical community that they use this situation to their advantage. Discussion is welcome!

  8. Is widespread “AI” integration inevitable? A foregone conclusion? I’m sorry to say that at least in the near term the answer is yes. Not because “resistance is futile” but because literally 97% of OECD countries have embedded AI into some aspect of their governance structure. It has ALREADY happened

    Opinion
  9. “These systems are solving problems in a few hours that I thought I’d be working on for years,” said a postdoctoral researcher in statistics. “It’s hard on the ego," Winer said. "But I’m much more concerned about the Mike-Winer-and-everyone-I-know-dying timeline.” www.nytimes.com/2026/11/22/m...

    Info
  10. I was sharing with my wife my (current) preferred approach to AI produced code by engineers. It's borrowed from her PhD - thesis defence. If an engineer can defend their code in a thesis defence, I have no concern about how it was produced. What is thesis defence for software changes? They should be able to explain, without using an llm: * What is the objective for the code change * Why did they choose this approach * What other approaches they considered * What unrelated/unexpected things they found along the way * What are the limitations of chosen approach * What is the wider/longer term impact of choosing this approach

    Show more of this post

    * Who else did they collaborate with, or who else understands their chosen approach to be able to answer these questions if needed Notice I don't want them to explain their work. I can read the code and understand it. I want to understand how well researched this approach is, and how well they understand it. I also want to ensure that they have put in at least as much effort (to produce the code or to prepare for the theses defence) as the reviewer will put in. #ai #softwareengineering

    Opinion
  11. @jtresser.bsky.social

    One of the most insightful pieces I’ve read on AI/ChatGPT. Why wouldn’t tech that relies on consuming intellectual labor to “evolve” reflect us - global humanity - and all our entrenched biases, systems, and power dynamics?

  12. @agowa338@chaos.social

    Today I tasked #Qwen with auditing the code it wrote. And it actually found a bunch of issues. (obviously not sure how much it missed). But I also don't know if it is funny or sad that the very same model that wrote the code can outline a bunch of security issues, including: * Token leak * Path traversal * Thread-safety * Caching issues * Missing limits on response sizes * XML injection in the nix module it wrote .... #AI #LLM

    Opinion
  13. Before Google existed, someone wrote down the fix for prompt tracking. It tells you which prompts belong on your list and what a right answer looks like. Thirty years later, search still hasn't picked it up. duaneforresterdecodes.substack.com/p/the-missin...

  14. @matdevdug@c.im

    As lukewarm as I am on #LLM, it has been interesting talking to friends about the death of #SAAS at their jobs. If a vendor raises their prices, there is now a real risk that a customer will simply use an LLM to copy the six features they want from that software and stop paying them all together. One took a Jira clone, added the two features they needed that the OSS version didn’t have and then bam end of contract. Took less than a week of work.

    Opinion
  15. I can't claim to have done any cutting-edge (or even buttering-edge) mathematics research, but the recent stories of LLM-based tools settling Navier-Stokes and Riemann Hypothesis have made me feel somewhat disoriented or dizzy. I feel oddly lightheaded, and maybe slightly depressed. But I'm not sad. I don't know how I'll feel if Google announces tomorrow that their AI system has proved that P is not equal to NP. None of these problems actually matter to me. I'm not a professional mathematician. I'm not an academic. On the other hand, I think my sensibilities are a bit academic.

    Opinion
  16. @ipolyzos.com

    Claude Code’s persistent permissions have exposed credentials through configuration files. A reminder that AI agent security extends beyond code: permissions, execution history, and configuration are part of the attack surface. www.infoq.com/news/2026/11... #AI #Security

    Info
  17. @coginiti.co

    Your AI agent just gave you a number because generating it was the easy part. Can you say whether the definition behind it was approved, tested, and current, and what produced it? A semantic layer alone can't. Semantic intelligence can, because trust lives below the surface.

  18. @grimalkina.bsky.social

    My thread the other day (on LLMs--> improving psych methods) has a mirror direction: can the methods of behavioral science/psychology inform how we study the "behavioral layer" of AI? www.science.org/doi/full/10....

  19. @masha@hoosier.social

    now imagine being a lawyer, where most of your job is based on legally tried and true boilerplate, except supervisors are asking you to rewrite boilerplate by plugging it into a model as a "template". and then they get frustrated when the output isn't close to identical. and the legal tech companies that piloted fill in form technology that makes "go-by" use more efficient are all using ai. *rests head softly on work laptop* my life would be so much easier if I also didn't understand LLMs

    Opinion
  20. @dzaladan.eurosky.social

    I'm a naturally curious person, so I like to explore any kind of topic. This thread here gives a precise explanation about how LLMs work. In a nutshell, confirms what has been said already: it's controlled randomization of preexisting data. So... math, not chemistry.

  21. @lena-tech.bsky.social

    AI-generated content may be cheap to produce, but it can be expensive for others to consume, verify and correct. A linguist’s survey of the vocabulary surrounding AI slop reveals how people perceive this transfer of costs. 1/7

  22. @erikahall.bsky.social

    The thing I keep coming back to is the degree to which automation is a fantasy. So many things aren't automated as much as they are orchestrated with capital, and easily disrupted or reallocated. And the below is the most distilled "cancer is our business model" elevator pitch I've ever seen.

    Opinion
  23. @paul4jags.bsky.social

    I see "not being a Luddite" is being worn as a badge of honour by several AI adopters who seem to misunderstand who the Luddites were and what they wanted. Here's a nice wee article from 2011, written by a human about them. Here's What the Luddites Really Fought Against share.google/hweGJwSTyMUi...

    Recommendation
  24. @jyasskin@hachyderm.io

    CEOs go off the deep end when they stop listening to their line workers, who actually know the underlying reality of the business. #LLMs offer everyone the chance to lose touch with reality in the same way, with workers that are programmed never to unionize or even to raise concerns unless you remember to explicitly prompt them to do so.

    Opinion
  25. @kwazekwaze.bsky.social

    In some respects, AI is an outgrowth of one of my least favorite aspects of internet culture: Sharing "content" without attribution and shutting down pushback as seeking attention. Apart from it being rude, it also gave us such hits as: "memes from avowed racists are super popular with kids"

    Opinion
  26. I think people should use AI in math as much as possible, but at the same time the new results have had no effect on the math I actually care about, where the goal is to understand things better and more deeply. Of course I am not familiar with every area of math, so researchers in those areas may gain some new insights from the AI results, but so far I have not seen anything that was applicable to anything I'm working on. It would be great to get an understanding of F_1 for example and prove the Riemann Hypothesis that way (and I like that Alpoge mentioned it), but just getting a proof of some prestigious conjecture doesn't impress me at all.

    Opinion
  27. @qlippot.bsky.social

    AI solved the Fermi Paradox, but not the way you think. Spacefaring civilizations don't exist to make contact with because sooner or later every sentient species puts its trust in charismatic grifters who piss away their planet's resources on glorified Skinner Boxes that they say will become AI.

    Opinion
  28. #claude #AI This disclaimer is not wrong. Describe me like you're trying to warn someone about me before they meet me: Don't be fooled by the calm guy from Aschaffenburg. He runs a tech team, so he looks harmless, but he's working on about eleven things at once and he will try to explain all of them to you. Don't mention printers. Don't show him a slow SQL query. He'll quietly turn your correlated subqueries into pre-aggregated CTEs and give it back running 40 times faster without saying a word. You'll feel grateful and a little embarrassed at the same moment.

    Show more of this post

    Don't say anything that could become a song. He has written death metal about inclusion, a Rolling Stones tribute apologizing to Keith Richards for the planet, a German garage-punk anthem about working from home without pants, and a power-metal hymn for sysadmins. Complain about traffic and you'll be a reggae-to-industrial track by Thursday. He also has an album series built on AI puns and plans to keep it going. Don't bring up dwarves, metals or rocks. You won't get away for an hour. He can sort ores by hardness, look into metallic glasses for a fantasy alloy, and lay out a forge's crafting loop while you're still finishing your coffee. Don't let him near your warehouse. He'll want to optimize it, and if he can't, he'll build a game about it. At the gaming table, he plays a Kor cleric with a tragic life debt and writes love songs for the other players' bards. He takes it seriously. Biggest danger: he usually spots your problem before you've finished explaining it, and the fix tends to involve PowerShell and a sea shanty. Verdict: It's safe to meet him, and you'll probably have a good time. Just block out more time than you planned. 🍺⚒️

    Creation
  29. @blackfyre.ninja

    Claude Code reads ~/.claude/CLAUDE.md in every session and repository, and I've published mine nearly in full. It opens with a glossary, as most rules depend on words like "assigned task" and "surface" meaning one thing, and three code words (Stop, Revert, Recap) override its autonomy at any point.

  30. @molly.wiki

    a lot of crypto companies use what i call the "buzzword shield", where they cram in jargon and hope that investors will assume whatever they do is too cutting-edge to understand but Datavault AI's "harnessing the power of supercomputing to subordinate AI" is a masterclass

    Popular todayprobably AI
  31. @humainlogic.bsky.social

    A new preprint (GAMA-Bench, 1,298 matched scenarios) found 10 leading AI models judged men more harshly than women in identical conflicts: more blame, less empathy, even with step-by-step reasoning. Tip: using AI to review complaints or HR issues? Swap the genders and compare. #AI #AIbias

    Info
  32. @marocddot@hachyderm.io

    Reading someone's code is like reading their thought processes. It's communication. It's knowing when to put in comments and when not to. It's putting fact and correctness into context. It's weaving abstractions that are enlightening (and even sometimes frustrating). Reading LLM produced code is dull and barren.

    Opinion
  33. @amkasprzyk.bsky.social

    AI companies “see the ability to write a rigorous proof of a mathematical fact as a crucial testing ground ... because the correctness of a proof is an objective measure that can be used as a ‘reward’ to further train AI systems.” It’s the wrong yardstick. 🧮 www.nature.com/articles/d41...

    Opinion
  34. @QNFO@mstdn.science

    Closing the loop: Ashby's ultrastable system and why today's AI systems disappoint A common belief: AI systems are smart enough already; what holds them back is model capability. What holds most AI systems back is an open loop: nothing checks the result against l [AI-generated]

    Info
  35. @parsingphase.dev

    A critical element in understanding the “AI boom” is to realize that its foremost proponents are walking incarnations of Brandolini’s Law - the principle that spewing bullshit is infinitely simpler than rebutting it. @adambecker.bsky.social’s More Everything forever does that work, thoroughly.

    Opinion
  36. @davidcrespo.bsky.social

    this is a great test and a good article but if you were to only read this post and the headline you might get a pretty wrong idea of what they found and what they recommend. all the issues were found in sketchy rando providers, some of which were wrongly trusted by OpenRouter

  37. @mathew@universeodon.com

    I keep telling people that you cannot trust any of the cloud AI services, they’re all run by thieves, liars, incompetents, and bad actors. This study adds to the evidence. Do not tell a chatbot anything you wouldn’t want posted on a public web page. To pick just one example, I did some research a while back and uncovered that BetterHelp is apparently feeding therapy transcripts to OpenAI. They don’t tell you, of course, not even in their privacy policy. I only found out because there was a press release elsewhere boasting about it.

  38. @ibboard@hachyderm.io

    Interesting views on AI contributions to open source projects from Miguel de Icaza. Personally, I'm not convinced by the "But we are still communicating with a human being on the other end" statement about the second wave of GenAI PRs. But I think the key bit is this: > Given that the economics of top-tier reasoning models aren't ideal for everyone doing casual open-source work, I’m happy to foot that bill on my end… We do need accountability. And a big way to do that is to give a shit, write the code and not use GenAI. tirania.org/blog/archive/2026/…

  39. @matthewboedy.bsky.social

    Well, folks, it’s been a fun ride: “He calculated about a 50 percent chance that humans would lose control of powerful A.I. systems in the next five years; 22 percent that most people would eventually die as a result...” www.nytimes.com/2026/11/22/m...

  40. @gilles@toot.gagniard.org

    That's to me the most impactful evolution in the field during the past few months. Thanks to progress in quantization methods and specifically-optimized inference engines such as Strata, it's now possible run relatively large MoE models such as Qwen 3.8 Flash Next (125B parameters) on regular PC. This is how we take back control. #LLM #qwen #strata #AI

  41. @ai-nerd.bsky.social

    slide 13 is where i stopped: RL took about 5% of the pretraining GPU-hours for DeepSeek R1 in Jan 2025, and more than 1.7x for Beam in Oct 2026. he notes GPU-hours overstate RL's share of FLOPs, since so much of RL is sampling

  42. @gabrielesvelto@mas.to

    This is a very interesting read and the only thing I'd like to add is about the part when he compares genAI usage in gamedev to things like YouTube enabling anyone to be a "filmmaker". We've seen this happen a number of times: having a tool doesn't magically make you a skilled artist, however all the previous times we've lowered the bar to do something we did it by making it *cheaper*. Digital audio recording, digital cameras, photo and video editing, digital painting, etc... In this case we're not. It'd be cheaper to offer free training for anyone who wants to get into gamedev than the torrent of money that is being spent on LLMs. Spending a lot to produce nothing of value. It doesn't get more ass-backwards than this. timeextension.com/news/2026/11…

    Opinion
  43. @popmediaprof.bsky.social

    troubling when this mentality trickles down/is absorbed by grad students. mentors are luddite elitists bc we expect scholars in training to *want* to read, think, know, engage in the processes of deep learning? (disagree re: the grinch. he got radicalized when they built a data center on his mtn)

    Opinion
  44. In addition to the proof of the Riemann hypothesis we now also have the proof that AI service providers resell your conversations (all of it), despite claims otherwise. What a time to be alive. arstechnica.com/security/2026/…

  45. @loleg@hachyderm.io

    A couple of teams made a point of implementing #LocalAI and one was quite excited to get #Apertus working in their financial advisor app 🫕 #powercoders #hackathon > We built TellHorn, a platform that integrates with your PostFinance app to bring real clarity to your spending, pairing your bank feed with receipt-level detail, and uses AI agents to actively optimize your expenses, spot hidden leaks, and find concrete ways to save money. github.com/TenzinJhopee/HackPo…

    Info
  46. @drhonor.bsky.social

    To characterize anti-AI sentiment as the “political left” is disingenuous and incorrect. It’s coming from young people who have been sold piles of dog shit by profit-minded elders for most of their lives and have developed razor-sharp bullshit detectors. This is generational, not left or right.

    Opinion
  47. Anthropic has cleverly managed to trick investors into throwing $100bn at pure mathematics and theoretical physics research. Gotta hand it to them, pretty cool. Imagine if we'd just gone the Bell Labs route and funded basic research with $100bn to see what sticks, maybe we'd also have been seeing huge breakthroughs of a similar magnitude, like we did last time we did that. And like and actually been able to understand and apply the research to. Weird idea, I know.

    OpinionPopular today
  48. @doug-hss.bsky.social

    AI is being insidiously integrated into universities by none other than Ewan Blair (Tony's son) via his company Multiverse, under the guise of workforce upskilling. Ultimately it's about workforce replacement while Multiverse makes millions in revenue via the government's apprenticeship levvy