• progress_activity cloud_sync

    Reconnection to the server…

    Movim cannot talk with the server, please try again later


    • Public subscriptions

    • chevron_right

      coopr8

      •
    • chevron_right

      gabagoo

      •
    • chevron_right

      kenu_demon

      •
    • chevron_right

      coopr8

      •
    • chevron_right

      gabagoo

      •
    • chevron_right

      kenu_demon

      •
    • chevron_right

      coopr8

      •
    • chevron_right

      gabagoo

      •
    • chevron_right

      kenu_demon

      •
  • Register Login

    Movim

    movim.chatterboxtown.us

  • group_work rss_feed
    add Follow

    ArsTechnica

    • Ar chevron_right

      Researchers concerned to find AI models hiding their true “reasoning” processes

      news.movim.eu / ArsTechnica • 10 April 2025

    Remember when teachers demanded that you "show your work" in school? Some fancy new AI models promise to do exactly that, but new research suggests that they sometimes hide their actual methods while fabricating elaborate explanations instead.

    New research from Anthropic—creator of the ChatGPT-like Claude AI assistant—examines simulated reasoning (SR) models like DeepSeek's R1 , and its own Claude series. In a research paper posted last week , Anthropic's Alignment Science team demonstrated that these SR models frequently fail to disclose when they've used external help or taken shortcuts, despite features designed to show their "reasoning" process.

    (It's worth noting that OpenAI's o1 and o3 series SR models deliberately obscure the accuracy of their "thought" process, so this study does not apply to them.)

    Read full article

    Comments

    • taglarge language models taglarge language models taglarge language models tagmachine learning tagmachine learning tagmachine learning tagsimulated reasoning tagsimulated reasoning tagsimulated reasoning tagsr models tagsr models tagsr models tagai tagai tagai tagbiz & it tagbiz & it tagbiz & it tagai alignment tagai alignment tagai alignment tagai research tagai research tagai research tagai safety tagai safety tagai safety taganthropic taganthropic taganthropic tagchatgpt tagchatgpt tagchatgpt tagclaude tagclaude tagclaude taglarge language models taglarge language models taglarge language models tagmachine learning tagmachine learning tagmachine learning tagsimulated reasoning tagsimulated reasoning tagsimulated reasoning tagsr models tagsr models tagsr models tagai tagai tagai tagbiz & it tagbiz & it tagbiz & it tagai alignment tagai alignment tagai alignment tagai research tagai research tagai research tagai safety tagai safety tagai safety taganthropic taganthropic taganthropic tagchatgpt tagchatgpt tagchatgpt tagclaude tagclaude tagclaude tagai tagai tagai tagbiz & it tagbiz & it tagbiz & it tagai alignment tagai alignment tagai alignment tagai research tagai research tagai research tagai safety tagai safety tagai safety taganthropic taganthropic taganthropic tagchatgpt tagchatgpt tagchatgpt tagclaude tagclaude tagclaude taglarge language models taglarge language models taglarge language models tagmachine learning tagmachine learning tagmachine learning tagsimulated reasoning tagsimulated reasoning tagsimulated reasoning tagsr models tagsr models tagsr models

    • Pictures 3 image

    • visibility
    • visibility
    • visibility
    • Ar chevron_right

      Researchers concerned to find AI models hiding their true “reasoning” processes

      news.movim.eu / ArsTechnica • 10 April 2025

    Remember when teachers demanded that you "show your work" in school? Some fancy new AI models promise to do exactly that, but new research suggests that they sometimes hide their actual methods while fabricating elaborate explanations instead.

    New research from Anthropic—creator of the ChatGPT-like Claude AI assistant—examines simulated reasoning (SR) models like DeepSeek's R1 , and its own Claude series. In a research paper posted last week , Anthropic's Alignment Science team demonstrated that these SR models frequently fail to disclose when they've used external help or taken shortcuts, despite features designed to show their "reasoning" process.

    (It's worth noting that OpenAI's o1 and o3 series SR models deliberately obscure the accuracy of their "thought" process, so this study does not apply to them.)

    Read full article

    Comments

    • tagai tagai tagai tagbiz & it tagbiz & it tagbiz & it tagai alignment tagai alignment tagai alignment tagai research tagai research tagai research tagai safety tagai safety tagai safety taganthropic taganthropic taganthropic tagchatgpt tagchatgpt tagchatgpt tagclaude tagclaude tagclaude tagai tagai tagai tagbiz & it tagbiz & it tagbiz & it tagai alignment tagai alignment tagai alignment tagai research tagai research tagai research tagai safety tagai safety tagai safety taganthropic taganthropic taganthropic tagchatgpt tagchatgpt tagchatgpt tagclaude tagclaude tagclaude taglarge language models taglarge language models taglarge language models tagmachine learning tagmachine learning tagmachine learning tagsimulated reasoning tagsimulated reasoning tagsimulated reasoning tagsr models tagsr models tagsr models taglarge language models taglarge language models taglarge language models tagmachine learning tagmachine learning tagmachine learning tagsimulated reasoning tagsimulated reasoning tagsimulated reasoning tagsr models tagsr models tagsr models tagai tagai tagai tagbiz & it tagbiz & it tagbiz & it tagai alignment tagai alignment tagai alignment tagai research tagai research tagai research tagai safety tagai safety tagai safety taganthropic taganthropic taganthropic tagchatgpt tagchatgpt tagchatgpt tagclaude tagclaude tagclaude taglarge language models taglarge language models taglarge language models tagmachine learning tagmachine learning tagmachine learning tagsimulated reasoning tagsimulated reasoning tagsimulated reasoning tagsr models tagsr models tagsr models

    • Pictures 3 image

    • visibility
    • visibility
    • visibility
    • Ar chevron_right

      Researchers concerned to find AI models hiding their true “reasoning” processes

      news.movim.eu / ArsTechnica • 10 April 2025

    Remember when teachers demanded that you "show your work" in school? Some fancy new AI models promise to do exactly that, but new research suggests that they sometimes hide their actual methods while fabricating elaborate explanations instead.

    New research from Anthropic—creator of the ChatGPT-like Claude AI assistant—examines simulated reasoning (SR) models like DeepSeek's R1 , and its own Claude series. In a research paper posted last week , Anthropic's Alignment Science team demonstrated that these SR models frequently fail to disclose when they've used external help or taken shortcuts, despite features designed to show their "reasoning" process.

    (It's worth noting that OpenAI's o1 and o3 series SR models deliberately obscure the accuracy of their "thought" process, so this study does not apply to them.)

    Read full article

    Comments

    • tagai tagai tagai tagbiz & it tagbiz & it tagbiz & it tagai alignment tagai alignment tagai alignment tagai research tagai research tagai research tagai safety tagai safety tagai safety taganthropic taganthropic taganthropic tagchatgpt tagchatgpt tagchatgpt tagclaude tagclaude tagclaude tagai tagai tagai tagbiz & it tagbiz & it tagbiz & it tagai alignment tagai alignment tagai alignment tagai research tagai research tagai research tagai safety tagai safety tagai safety taganthropic taganthropic taganthropic tagchatgpt tagchatgpt tagchatgpt tagclaude tagclaude tagclaude taglarge language models taglarge language models taglarge language models tagmachine learning tagmachine learning tagmachine learning tagsimulated reasoning tagsimulated reasoning tagsimulated reasoning tagsr models tagsr models tagsr models tagai tagai tagai tagbiz & it tagbiz & it tagbiz & it tagai alignment tagai alignment tagai alignment tagai research tagai research tagai research tagai safety tagai safety tagai safety taganthropic taganthropic taganthropic tagchatgpt tagchatgpt tagchatgpt tagclaude tagclaude tagclaude taglarge language models taglarge language models taglarge language models tagmachine learning tagmachine learning tagmachine learning tagsimulated reasoning tagsimulated reasoning tagsimulated reasoning tagsr models tagsr models tagsr models taglarge language models taglarge language models taglarge language models tagmachine learning tagmachine learning tagmachine learning tagsimulated reasoning tagsimulated reasoning tagsimulated reasoning tagsr models tagsr models tagsr models

    • Pictures 3 image

    • visibility
    • visibility
    • visibility
  • cloud_queue

    Powered by Movim