行业多源确认
Bindu Reddy 质疑 Anthropic 关于小型模型自我复制的说法
精选理由
Bindu Reddy 吐槽 Anthropic 说小模型能自我复制，她觉得网上几百万个小模型都能联网，也没见出什么事，看看双方怎么看 AI 风险。
Airtable 前高管 Bindu Reddy 在 X 上发文，质疑 Anthropic 声称小型 AI 模型可以自我复制的观点。她认为小型模型本身无害，难以构成对 NSA 这类机构的实际威胁。她进一步指出，互联网上已有数百万个小型模型，且大多数可以直接访问网络。这篇帖子属于对 Anthropic AI 风险评估立场的公开评论。
原文 · Bindu Reddy
LOL, Anthropic seems pretty determined to prove that AI is an existential threat
Now they claim small AI models can replicate themselves 🤷♀️
Ok, but small AI models are kinda harmless and can’t break into NSA
In fact, we have millions of them lying around on the internet today - most of them with direct access to the internet
