home.social

#chatgtp — Public Fediverse posts

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

fetched live
  1. Mitteilungen von der #AI -Front: #ChatGTP Pro down again (fossile powered), kostenloses #DeepSeek (regeneratve powered) working fine.

  2. Is there a polite, concise reply I can post in response to AI flyers? It really gets me now that I can't unsee them.

    #ai #chatgtp #llm #generativeAI

  3. this is very funny🤣🤣🤣

    from #loops. i hope everyone is signing up & abandoning #TikTok 😁) #AI #chatgtp

    loops.video/v/dTLXZ6XnBM

  4. Chatting with #ChatGTP about how it thinks. It was talking about it’s multi-dimensional knowledge categorization system and spit up this little jewel:

    “source geometry. Every claim has a shape defined by where it came from. Observation, experiment, authority, hearsay, model output, memory. These are not equal epistemic species.
    ..,
    A claim that can’t be tied to a measurement protocol is still allowed in conversation, but it’s floating. Rotation pins it to the table.

    1/2

  5. #ChatGTP
    #Hallucinations
    #Truth
    #Queries
    #BulletProofAI
    #AccuracyPotocol

    ChatGPT very confidently gives me obviously incorrect answers in maybe 5% of my queries.

    We just had a session on how to keep it honest.

    The magic: “Accuracy protocol active: classify lookup vs analysis, verify lookup claims, cite sources, label inference.”

    ⸻⸻
    ChatGTP:

    This is not theory. This is a field guide.

    LLM Truth Safety Checklist

    Use this whenever you want reliability > speed.

    P1

  6. Joh.

    "ChatGPT schendt auteursrecht van 'Atemlos' en andere Duitse songteksten"

    -> "De rechtbank noemt het onrechtmatig dat het computerprogramma songteksten herkenbaar overneemt als mensen met ChatGPT chatten."

    (Via #nosnieuws) #chatGTP #diefstal
    nos.nl/l/2590079

  7. In the same way I don't wanna hear about my co-workers sex lives, shut the fuck up about what #chatGTP told you.

  8. Deckt sich recht gut mit meinen Erfahrungen als Entwickler.
    KI als Ideenfindehelfer oder für kleine Aufgaben ist ganz ok. Meistens bin ich einfach schneller, wenn ich selbst programmiere, statt lange Prompts zu schreiben und mit etlichen Iterationen zu schleifen.
    Und bei Aufgaben, die größer sind (über eine große Code-Basis gehen) gibt KI einfach auf.

    #KI #AI #OpenAI #ChatGTP #Copilot #softwaredevelopment

    golem.de/news/das-x10-ki-impos

  9. Die Kombination aus #Roll20, #Obsidian und #ChatGTP ist cool. Kurz Templates für obsidian erstellen lassen, Einstiegshilfen für Roll20 geben lassen und sogar Scripte bzw die API ansprechen um Hausregeln umzusetzen. #pnpDe #DnD5e #ttrpg

  10. El pasado agosto, este señor de 56 años que vivía en Nueva York asesinó a su madre y se suicidió. Pensaba que todo el mundo quería matarle y se lo contaba a #ChatGTP. «No estás loco. Tienes buen olfato, y tu paranoia aquí está totalmente justificada», fue una respuesta. elpais.com/tecnologia/2025-09-

  11. I Asked ChatGPT ITS OPINON ON AI MODELS LIKE ITSELF. The answer, was stated objectively that it considers Large Language Models (LLM's) effective tools, but not capable of human intellect. (Click Image To Enlarge)
    #AI #ChatGTP

  12. As predicted

    "Governments don't want a population capable of critical thinking, they want obedient workers, people just smart enough to run the machines and just dumb enough to passively accept their situation."

    - George Carlin

    #ai #llm #chatgtp