home.social

#lecun — Public Fediverse posts

Live and recent posts from across the Fediverse tagged #lecun, aggregated by home.social.

  1. @blog defence-in-depth is still the most rational stratagem and it is cheaper than "AI" but requires architecting software correctly from the get-go with domain and process knowledge, the very #epistemological nature of which is largely alien to #LLM driven automation approaches. in other words, it requires human judgement by its very nature. the agentic approaches are imitative only, the LLMs do not and cannot "understand". #Lecun knows this as do #Hinton, yet #Amodei threw us all under a bus.

  2. @blog defence-in-depth is still the most rational stratagem and it is cheaper than "AI" but requires architecting software correctly from the get-go with domain and process knowledge, the very #epistemological nature of which is largely alien to #LLM driven automation approaches. in other words, it requires human judgement by its very nature. the agentic approaches are imitative only, the LLMs do not and cannot "understand". #Lecun knows this as do #Hinton, yet #Amodei threw us all under a bus.

  3. @blog defence-in-depth is still the most rational stratagem and it is cheaper than "AI" but requires architecting software correctly from the get-go with domain and process knowledge, the very #epistemological nature of which is largely alien to #LLM driven automation approaches. in other words, it requires human judgement by its very nature. the agentic approaches are imitative only, the LLMs do not and cannot "understand". #Lecun knows this as do #Hinton, yet #Amodei threw us all under a bus.

  4. @blog defence-in-depth is still the most rational stratagem and it is cheaper than "AI" but requires architecting software correctly from the get-go with domain and process knowledge, the very #epistemological nature of which is largely alien to #LLM driven automation approaches. in other words, it requires human judgement by its very nature. the agentic approaches are imitative only, the LLMs do not and cannot "understand". #Lecun knows this as do #Hinton, yet #Amodei threw us all under a bus.

  5. @blog defence-in-depth is still the most rational stratagem and it is cheaper than "AI" but requires architecting software correctly from the get-go with domain and process knowledge, the very #epistemological nature of which is largely alien to #LLM driven automation approaches. in other words, it requires human judgement by its very nature. the agentic approaches are imitative only, the LLMs do not and cannot "understand". #Lecun knows this as do #Hinton, yet #Amodei threw us all under a bus.

  6. LeCun gab der US-Ausgabe von MIT Technology Review im Januar in einem Interview aus seiner Pariser Wohnung exklusive Einblicke. Er spricht über sein neues Unternehmen, sein Leben nach Meta, die Zukunft der künstlichen Intelligenz. (+)

    #LeCun #Meta #KI #Ami

    t3n.de/news/meta%e2%80%91ki-pi

  7. LeCun gab der US-Ausgabe von MIT Technology Review im Januar in einem Interview aus seiner Pariser Wohnung exklusive Einblicke. Er spricht über sein neues Unternehmen, sein Leben nach Meta, die Zukunft der künstlichen Intelligenz. (+)

    #LeCun #Meta #KI #Ami

    t3n.de/news/meta%e2%80%91ki-pi

  8. #YannLeCun, a leading #AI #researcher, believes the current focus on #LLMs is misguided and will not solve many pressing problems. He advocates for #worldmodels, a different type of AI that accurately reflects the real world, and is a strong proponent of #opensource AI. #LeCun’s new company, #AMI Labs, will focus on developing world models. technologyreview.com/2026/01/2 #tech #media #news

  9. In this interview with Yann LeCun in MIT's Technology Review he explains why his new company AMI (pronounced 'Amie' in French) steps away from LLMs.

    'The reason why a 17-year-old can learn to drive in 20 hrs is that they already know a lot about how the world behaves.

    If we want a generally useful domestic robot, we need systems to have a good understanding of the physical world. That’s not going to happen until we have good world models and planning.'

    #LeCun #AMI #WorldModels #AI
    @dogzilla

  10. LeCun verlässt Meta für eigenes Startup: LLMs sind am Ende:

    - Rücktritt Ende 2025 für volle Autonomie ohne Produktzwang
    - Abkehr von stochastischen Papageien hin zu Weltmodellen
    - AMI statt LLM: Fokus auf echte Planung und Kausalität

    Ist die aktuelle Transformer-Architektur wirklich eine technische Sackgasse? #LeCun #Meta #AMI
    all-ai.de/news/news24/lecun-me

  11. @arstechnica Pour ceux qui comprennent le français, il y a de cela deux semaines, Yann LeCun a fait une leçon magistrale avec une craie et un tableau de classe dans laquelle il décrit non seulement les algorithmes des grands modèles de langage (LLM), mais aussi les modèles du monde (world models) qui attirent maintenant davantage son attention et ses efforts de recherche.

    youtube.com/watch?v=Z208NMP7_-0

    #AI #ArtificialIntelligence #LeCun #LLM #WorldModel

  12. Große Sprachmodelle wie GPT-4 sind beeindruckend – aber laut Yann LeCun nur clevere "Hacks". Auf dem AI Action Summit stellt er mit V-JEPA einen grundlegend anderen Ansatz vor. Was fehlt aktueller KI, und was plant Meta wirklich? Jetzt mehr erfahren. #Meta #LeCun #KI 👇
    all-ai.de/news/topbeitraege/le