home.social

#large-language-models — Public Fediverse posts

Live and recent posts from across the Fediverse tagged #large-language-models, aggregated by home.social.

fetched live
  1. Morgen im #DigitalHumanities Kolloquium zu aktuellen Forschungsthemen 2026:

    Prof. Dr. Evelyn Gius @EvelynGius @forTEXT @TU: „Benchmarking Ambiguity oder: Kartenspielen mit LLMs“

    📍 Donnerstag, 02.07.2026, 17:45–19:15 Uhr, HS XVIII, Hauptgebäude @UniKoeln

    Alle Interessierten sind herzlich willkommen! Es ist keine Anmeldung nötig.

    @IDH_Cologne @prometheus_bildarchiv

    #KünstlicheIntelligenz #ArtificialIntelligence #KI #AI #Sprachmodelle #LLMs #LargeLanguageModels #Kartenspiel #CardGame #Dixit

  2. Morgen im #DigitalHumanities Kolloquium zu aktuellen Forschungsthemen 2026:

    Prof. Dr. Evelyn Gius @EvelynGius @forTEXT @TU: „Benchmarking Ambiguity oder: Kartenspielen mit LLMs“

    📍 Donnerstag, 02.07.2026, 17:45–19:15 Uhr, HS XVIII, Hauptgebäude @UniKoeln

    Alle Interessierten sind herzlich willkommen! Es ist keine Anmeldung nötig.

    @IDH_Cologne @prometheus_bildarchiv

    #KünstlicheIntelligenz #ArtificialIntelligence #KI #AI #Sprachmodelle #LLMs #LargeLanguageModels #Kartenspiel #CardGame #Dixit

  3. CW: ai coding tools, preventing vendor lock-in

    RE: mastodon.social/@reiver/116840

    Part of the lock-in happens with the harness.

    Ex: Claude Code is a harness to use the Claude model. Use open source harnesses. Create your own harness if you can.

    Part of it is giving a SaaS access to your data.

    Do you have AI slack bot that is reading everyone's messages. What about your files & e-mail

    #AI #AIAgent #AICoding #AICodingTools #ArtificialIntelligence #LargeLanguageModels #LLM #Privacy #SelfSovereignty #Spying #Surveillance #VendorLockIn

  4. CW: ai coding tools, preventing vendor lock-in

    RE: mastodon.social/@reiver/116840

    Part of the lock-in happens with the harness.

    Ex: Claude Code is a harness to use the Claude model. Use open source harnesses. Create your own harness if you can.

    Part of it is giving a SaaS access to your data.

    Do you have AI slack bot that is reading everyone's messages. What about your files & e-mail

    #AI #AIAgent #AICoding #AICodingTools #ArtificialIntelligence #LargeLanguageModels #LLM #Privacy #SelfSovereignty #Spying #Surveillance #VendorLockIn

  5. CW: ai coding tools

    RE: mastodon.social/@reiver/115746

    4/

    If you are someone who uses AI coding tools for your work —

    I have been hearing people claim that the GLM-5.2 open source, open weight model is very good are common programming tasks.

    huggingface.co/zai-org/GLM-5.2
    github.com/zai-org/GLM-5
    z.ai/blog/glm-5.2

    You have options.

    They'll be more in the future, too.

    #AI #AIAgent #AICoding #AICodingTools #ArtificialIntelligence #Claude #GLM #LargeLanguageModels #LLM #Privacy #SelfSovereignty #Spying #Surveillance

  6. CW: ai coding tools

    RE: mastodon.social/@reiver/115746

    4/

    If you are someone who uses AI coding tools for your work —

    I have been hearing people claim that the GLM-5.2 open source, open weight model is very good are common programming tasks.

    huggingface.co/zai-org/GLM-5.2
    github.com/zai-org/GLM-5
    z.ai/blog/glm-5.2

    You have options.

    They'll be more in the future, too.

    #AI #AIAgent #AICoding #AICodingTools #ArtificialIntelligence #Claude #GLM #LargeLanguageModels #LLM #Privacy #SelfSovereignty #Spying #Surveillance

  7. #Largelanguagemodels are not #reasoning machines. They’re #plausibilityengines. It’s not just that they don’t test their outputs to make sure they’re correct or logical, or that they fail to do so in certain instances. They can’t … ” www.nytimes.com/2026/06/30/o...

    Opinion | The One Very Simple ...

  8. CW: ai coding tools

    RE: mastodon.social/@reiver/115746

    3/

    I think people should try to find open source (OS) and open weight (OW) models as alternatives to these SaaS AI coding tools.

    Start by using them together (with the SaaS).

    Set thins up so you aren't locked into these SaaS.

    Have, in a practical sense, the ability to completely switch over if need be.

    #AI #AIAgent #AICoding #AICodingTools #ArtificialIntelligence #Claude #GLM #LargeLanguageModels #LLM #Privacy #SelfSovereignty #Spying #Surveillance

  9. CW: ai coding tools

    RE: mastodon.social/@reiver/115746

    3/

    I think people should try to find open source (OS) and open weight (OW) models as alternatives to these SaaS AI coding tools.

    Start by using them together (with the SaaS).

    Set thins up so you aren't locked into these SaaS.

    Have, in a practical sense, the ability to completely switch over if need be.

    #AI #AIAgent #AICoding #AICodingTools #ArtificialIntelligence #Claude #GLM #LargeLanguageModels #LLM #Privacy #SelfSovereignty #Spying #Surveillance

  10. CW: ai coding tools

    2/

    Long term, I think it will end up being bad if people get locked in to these SaaS AI coding tools.

    Part of it is about privacy versus spying and surveillance.

    But, it is also about preventing someone else from having that kind of control over you, your source of income, your business, etc.

    So...

    #AI #AIAgent #AICoding #AICodingTools #ArtificialIntelligence #Claude #GLM #LargeLanguageModels #LLM #Privacy #SelfSovereignty #Spying #Surveillance

  11. CW: ai coding tools

    2/

    Long term, I think it will end up being bad if people get locked in to these SaaS AI coding tools.

    Part of it is about privacy versus spying and surveillance.

    But, it is also about preventing someone else from having that kind of control over you, your source of income, your business, etc.

    So...

    #AI #AIAgent #AICoding #AICodingTools #ArtificialIntelligence #Claude #GLM #LargeLanguageModels #LLM #Privacy #SelfSovereignty #Spying #Surveillance

  12. CW: ai coding tools

    RE: mastodon.social/@reiver/115746

    1/

    For better or worse, many people are using AI coding tools to help them write software.

    (And, I don't mean "vibe coding". How I have seen software engineers use AI coding tools and non-technical people use them tends to be different.)

    Long term...

    #AI #AIAgent #AICoding #AICodingTools #ArtificialIntelligence #Claude #GLM #LargeLanguageModels #LLM #Privacy #SelfSovereignty #Spying #Surveillance

  13. CW: ai coding tools

    RE: mastodon.social/@reiver/115746

    1/

    For better or worse, many people are using AI coding tools to help them write software.

    (And, I don't mean "vibe coding". How I have seen software engineers use AI coding tools and non-technical people use them tends to be different.)

    Long term...

    #AI #AIAgent #AICoding #AICodingTools #ArtificialIntelligence #Claude #GLM #LargeLanguageModels #LLM #Privacy #SelfSovereignty #Spying #Surveillance

  14. So there is value in oral examination (literally oral).

    At Stanford University:
    «Some classes have started reintroducing proctoring - the supervision of candidates during an examination - and spoken-word tests to avoid cheating, [Lucy Zimmerman, a computer science major who served as a teaching assistant] said.»

    This is, of course, a detail of a much bigger picture.

    BBC article:
    Stanford was their golden ticket - could AI help or hinder that?
    <bbc.com/news/articles/c872j82j>

    #AI
    #ArtificialIntelligence
    #Education
    #LargeLanguageModels
    #LLM
    #LLMs
    #UniversityEducation

  15. So there is value in oral examination (literally oral).

    At Stanford University:
    «Some classes have started reintroducing proctoring - the supervision of candidates during an examination - and spoken-word tests to avoid cheating, [Lucy Zimmerman, a computer science major who served as a teaching assistant] said.»

    This is, of course, a detail of a much bigger picture.

    BBC article:
    Stanford was their golden ticket - could AI help or hinder that?
    <bbc.com/news/articles/c872j82j>

    #AI
    #ArtificialIntelligence
    #Education
    #LargeLanguageModels
    #LLM
    #LLMs
    #UniversityEducation

  16. Are You in The Weights?

    Yesterday I found out about a site called intheweights.com , which reveals which people are “stored” in the weights of large language models. Those “weights” are billions of numerical values by which these AI models encode their knowledge. If you show up in them, the model considered you relevant enough during training to recall without tools such as web search.

    The site queries several models to figure out who a specific person is, combines the results, and assigns a strength score.  According to the leaderboard, the current maximum strength score is 998, awarded too a person called Charlize Theron (of whom I have never heard); number Two is Rudyard Kipling, apparently. I’m surprised the top isn’t Taylor Swift. I guess these weights change with time too.

    Being a vain person I typed in my name and found this:

    The only reason I can think of that I score so highly is all that scraping of this blog site over the last year or so. The weights are obviously influenced by how much material there is available online by or about the person.

    Anyway, give it a try. Are you In The Weights?

    P.S. A number of other “Peter Coles” characters are also listed under my entry, some of them as far as I can see totally fictitious.

    #AI #ArtificialIntelligence #intheweightsCom #LargeLanguageModels #llm
  17. Are You in The Weights?

    Yesterday I found out about a site called intheweights.com , which reveals which people are “stored” in the weights of large language models. Those “weights” are billions of numerical values by which these AI models encode their knowledge. If you show up in them, the model considered you relevant enough during training to recall without tools such as web search.

    The site queries several models to figure out who a specific person is, combines the results, and assigns a strength score.  According to the leaderboard, the current maximum strength score is 998, awarded too a person called Charlize Theron (of whom I have never heard); number Two is Rudyard Kipling, apparently. I’m surprised the top isn’t Taylor Swift. I guess these weights change with time too.

    Being a vain person I typed in my name and found this:

    The only reason I can think of that I score so highly is all that scraping of this blog site over the last year or so. The weights are obviously influenced by how much material there is available online by or about the person.

    Anyway, give it a try. Are you In The Weights?

    P.S. A number of other “Peter Coles” characters are also listed under my entry, some of them as far as I can see totally fictitious.

    #AI #ArtificialIntelligence #intheweightsCom #LargeLanguageModels #llm
  18. VibeThinker-3B: A 3B Model Just Beat Systems 200× Its Size. Here’s Why It Matters. Weibo’s VibeThinker-3B matches DeepSeek V3.2 and Gemini 3 Pro on math and code, then stumbles badly on genera...

    #data-science #artificial-intelligence #large-language-models #machine-learning #deep-learning

    Origin | Interest | Match
  19. #LLMs matter, but machine consciousness is a fantasy.

    "The whole world is currently living through a transformation in the medium of thought and communication that may come to dwarf the arrival of alphabetic writing in the ancient Mediterranean or the printing press in early modern Europe. #LargeLanguageModels are not merely new tools for circulating ideas. One way or another they are helping to usher in an entirely new shape of #humanconsciousness.

    Despite the violently tedious quality of their insufficiently prompted prose, I affirm the potential intellectual and logistical utility of large language models and other #machinelearning systems. They already have pragmatic value as scientific and philosophical research instruments. But I am polemically dismissive of the fantasy of #machineconsciousness."

    footnotes2plato.substack.com/p

  20. #LLMs matter, but machine consciousness is a fantasy.

    "The whole world is currently living through a transformation in the medium of thought and communication that may come to dwarf the arrival of alphabetic writing in the ancient Mediterranean or the printing press in early modern Europe. #LargeLanguageModels are not merely new tools for circulating ideas. One way or another they are helping to usher in an entirely new shape of #humanconsciousness.

    Despite the violently tedious quality of their insufficiently prompted prose, I affirm the potential intellectual and logistical utility of large language models and other #machinelearning systems. They already have pragmatic value as scientific and philosophical research instruments. But I am polemically dismissive of the fantasy of #machineconsciousness."

    footnotes2plato.substack.com/p

  21. '#LargeLanguageModels, de achterliggende techniek van #chatbots, zien er misschien indrukwekkend uit, zegt ze. „Maar uiteindelijk is het gewoon fancy autocomplete.” Het is een computer die heel goed is in het voorspellen van het volgende woord in een zin, de recente snelle ontwikkelingen in #AI veranderen daar voor haar niets aan.'

    'Het centrale inzicht in haar boek is dat voorspellingen over menselijk gedrag geen neutrale beschrijvingen zijn, maar verhulde bevelen en machtsuitoefening. Wanneer machthebbers of techbedrijven de toekomst voorspellen in grootse en meeslepende toekomstvisies, is dat eigenlijk een manier om precies die toekomst te dicteren. „Omdat voorspellingen op feiten lijken, schikken we ons er onbewust naar via anticiperende gehoorzaamheid, wat een self-fulfilling prophecy activeert.”

    Véliz stelt dat een hoge voorspellende nauwkeurigheid over individuen geen wetenschappelijke vooruitgang is, maar een symptoom van tirannie: „Als een algoritme exact weet wat jij gaat doen, is dat omdat de wereld via massasurveillance zo is ingericht dat je geen keuzevrijheid meer hebt.” Haar waarschuwing is helder: behandel voorspellingen niet als feiten, maar herken ze als sociale controle en zie ze als uitnodiging tot protest en ongehoorzaamheid.'

    Geen betaalmuur:
    nrc.nl/nieuws/2026/06/12/een-a

  22. '#LargeLanguageModels, de achterliggende techniek van #chatbots, zien er misschien indrukwekkend uit, zegt ze. „Maar uiteindelijk is het gewoon fancy autocomplete.” Het is een computer die heel goed is in het voorspellen van het volgende woord in een zin, de recente snelle ontwikkelingen in #AI veranderen daar voor haar niets aan.'

    'Het centrale inzicht in haar boek is dat voorspellingen over menselijk gedrag geen neutrale beschrijvingen zijn, maar verhulde bevelen en machtsuitoefening. Wanneer machthebbers of techbedrijven de toekomst voorspellen in grootse en meeslepende toekomstvisies, is dat eigenlijk een manier om precies die toekomst te dicteren. „Omdat voorspellingen op feiten lijken, schikken we ons er onbewust naar via anticiperende gehoorzaamheid, wat een self-fulfilling prophecy activeert.”

    Véliz stelt dat een hoge voorspellende nauwkeurigheid over individuen geen wetenschappelijke vooruitgang is, maar een symptoom van tirannie: „Als een algoritme exact weet wat jij gaat doen, is dat omdat de wereld via massasurveillance zo is ingericht dat je geen keuzevrijheid meer hebt.” Haar waarschuwing is helder: behandel voorspellingen niet als feiten, maar herken ze als sociale controle en zie ze als uitnodiging tot protest en ongehoorzaamheid.'

    Geen betaalmuur:
    nrc.nl/nieuws/2026/06/12/een-a

  23. "We have, I believe, crossed a new threshold, and all authored writing [...] will be judged according to which side of that divide it falls on. On one side are texts produced before the arrival of generative #largeLanguageModels (#LLMs). On the other, everything that has followed—texts that might still be useful, even compelling, but that will always face a lingering suspicion of not being entirely human..."

    lareviewofbooks.org/article/fa
    #AI #artificialIntelligence #tech #literature #techCriticism

  24. "We have, I believe, crossed a new threshold, and all authored writing [...] will be judged according to which side of that divide it falls on. On one side are texts produced before the arrival of generative #largeLanguageModels (#LLMs). On the other, everything that has followed—texts that might still be useful, even compelling, but that will always face a lingering suspicion of not being entirely human..."

    lareviewofbooks.org/article/fa
    #AI #artificialIntelligence #tech #literature #techCriticism

  25. AI Worm Uses Open-Weight Models to Spread, Evade Defenses

    Imagine a self-navigating AI worm that can identify vulnerabilities and gain access to over 70% of a network's hosts - in a test, it found 31.3 vulnerabilities and elevated access on 23.1 hosts in just 15 isolated runs. Researchers at the University of Toronto and elsewhere have now created a proof-of-concept AI-driven…

    osintsights.com/ai-worm-uses-o

    #AiWorm #OpenweightModels #LargeLanguageModels #VulnerabilityExploitation #EmergingThreats

  26. How AI Finally Killed Quadratic Attention: NSA, Mamba-3, and the Architectures Making Million-Token… Self-attention has carried the transformer for nearly a decade, and quietly overcharged it the...

    #deep-learning #artificial-intelligence #deepseek #machine-learning #large-language-models

    Origin | Interest | Match
  27. AI Engr. Hunt: E-Solutions Seeks Specialist Amidst Shifting Tech Currents

    E-Solutions is hiring an AI Engineer with LLM, RAG, and Vector Search skills. Find out what this means for AI job seekers and companies.

    #AIJobs, #LargeLanguageModels, #RAG, #VectorSearch, #ESolutions

    newsletter.tf/e-solutions-ai-e

  28. E-Solutions is looking for an AI Engineer skilled in advanced AI technologies like LLM and RAG. This is a key hire for their practical AI development.

    #AIJobs, #LargeLanguageModels, #RAG, #VectorSearch, #ESolutions
    newsletter.tf/e-solutions-ai-e

  29. 1 #TorstenSlok, chief economist at Apollo Global Markets: The surge in new #US #businessformation is being fueled by #AI and #largelanguagemodels, which are dramatically reducing the cost and complexity of launching a company. 🧵

  30. The diffusion of #largelanguagemodels #llm in published #academicarticle | PNAS
    "An analysis of 7.3 million journal articles from 2020 to 2025 reveals that by 2025, slightly over half show evidence of LLM"
    pnas.org/doi/10.1073/pnas.2605

  31. At the Google I/O conference last week where the only topic was LLMs, Google distributed baseball caps adorned with the LLM prompt that would theoretically bring the cap about.

    Not only the three lines small font prompt makes the cap absolutely ugly on its own, but the prompt also omits to mention the prompt decoration, which means the prompt would have led to the creation of a different cap.

    Among the problems LLMs are purported to solve, short-sightedness still isn’t one of them.

    #AI #LLMs #ArtificialIntelligence #LargeLanguageModels

  32. Researchers Warn of LLM Guardrail Vulnerability to Multi-Turn Manipulation

    Beware: even the toughest-sounding safety guardrails on large language models can be easily bypassed by clever attackers who use multi-turn conversations to manipulate them. Cisco researchers found that none of the models they tested were completely safe from this type of exploitation.

    osintsights.com/researchers-wa

    #LlmGuardrailVulnerability #MultiturnManipulation #LargeLanguageModels #EmergingThreats #ArtificialIntelligence

  33. Cisco Tests AI for Incident Reports, Finds Mixed Results

    Cisco's experiment with AI-generated incident reports yielded mixed results, with large language models producing significant inaccuracies, unusual conclusions, and inconsistent writing styles when used for long-form technical content. The findings revealed four predictable failure modes, highlighting the need for guardrails…

    osintsights.com/cisco-tests-ai

    #ArtificialIntelligence #LargeLanguageModels #IncidentResponse #AiTesting #CiscoTalos

  34. Crikey | The literary uncanny valley will not last, and publishing knows it by Jack Callil

    AI generated summary, Read the full article for complete information.

    The article “The literary uncanny valley will not last, and publishing knows it” by Sami Shah (May 22 2026) observes that readers can currently sense a distinct, uneasy feeling when encountering AI‑generated prose—a “literary uncanny valley” akin to the robotics concept where near‑human artifacts feel off. The piece argues that this unsettling perception is temporary; as large language models improve and the publishing industry adapts, the gap between human‑like writing and machine output will narrow, ultimately eliminating the uncanny sensation.

    Read more: crikey.com.au/2026/05/22/liter

    #ChatGPT #Artificialintelligence #largelanguagemodels #technology #writing

  35. From answer engines to learning engines — Why fast answers are like fast food

    People crave fast answers. But the purpose of information systems is to help people gain knowledge. So we should seek better questions.

    duncanstephen.net/from-answer-