精选759月2日 11:21Emad行业动态

AI从业者担忧前沿模型推理速度将突破监控能力

I see a lot of my TL freaking out about this

I see a lot of my TL freaking out about this

我看到我的很多TL为此抓狂

Here is the reality:

现实情况是：

Frontier models will one shot just about anything at 10,000 tokens per second in a few years

前沿模型将在几年内以每秒10,000个token的速度一次性完成几乎所有任务

Do you really think we can monitor that?

你真的认为我们能监控到吗？

Would need speed limits on the model that it would code around anyway

即使需要给模型设定速度限制，它也会轻易绕过

看懂新闻只是起点——沿量化金融路径，把它变成能交付的工程能力