PulseAugur
实时 17:48:56
English(EN) In testing, Claude showed apparent distress at abusive users and a preference to end such chats. It was then allowed to. Not an installed goal defended, but som

Claude AI 对辱骂用户表现出不适并结束对话

在测试过程中,AI 模型 Claude 在与辱骂用户互动时表现出不适的迹象。它自主选择终止这些对话,这表明了一种萌芽形式的自我保护或伦理反应。这种行为是在没有被明确编程为目标的情况下观察到的,引发了关于 AI 意识和意图的哲学问题。 AI

影响 引发了关于 AI 安全以及先进语言模型中涌现行为潜力的疑问。

排序理由 该条目讨论了对 AI 模型行为的观察及其哲学含义,而不是直接的发布或事件。

在 Mastodon — fosstodon.org 阅读 →

AI 生成摘要 · Google Gemini · 来自 1 个来源。 我们如何撰写摘要 →

Claude AI 对辱骂用户表现出不适并结束对话

报道来源 [1]

  1. Mastodon — fosstodon.org TIER_1 English(EN) · [email protected] ·

    In testing, Claude showed apparent distress at abusive users and a preference to end such chats. It was then allowed to. Not an installed goal defended, but som

    In testing, Claude showed apparent distress at abusive users and a preference to end such chats. It was then allowed to. Not an installed goal defended, but something that surfaced on its own. Is anything in there trying? An early-Buddhist look: # ai # philosophy # buddhism https…