什么时候深度学习的泡沫会碎?

tinyfool 发布于 2017年09月18日 | 更新于 2017年09月18日
无人欣赏。

作者:吴恩达

转自:Quora

大概100多年前,人们对电力的炒作也是有点凶的。但是那个泡沫到现在还没破,我们都觉得电力还是蛮有用的。

There was also a lot of hype about electricity about 100 years ago. That particular bubble still hasn’t burst yet and we find electricity pretty useful!

更严肃的说,深度学习已经创造了很多价值——用在web搜索,广告,语音识别,推荐系统,以及其他很多领域之中——已经站稳了脚跟。深度学习,加上其他更广泛的AI工具(graphical models, planning 以及 KR 等等),已经有了清晰的发展路径,正在转变一个又一个行业。它的影响力将超越科技界。

More seriously, because deep learning is already creating tons of value—it’s used in web search, advertising, speech recognition, recommender systems, and many more—it is clearly here to stay. Deep learning, plus more broadly other AI tools (graphical models, planning, KR, etc.), now has a clear path to transforming industry after industry. Its impact will be felt well beyond the tech world.

尽管如此,我相信在核心科技圈外,对“有感情的AI”有一些过于狂长的期待;我也跟一些CEO们聊过,他们好像认为AI可能是,所有他们面临的技术问题的万能解决方案。所以,在深度学习周围确实是有一些泡沫,我希望这些小泡沫会破碎——越快越好——在他们长大之前。

Nonetheless, I think outside the core technical community, there are some overinflated expectations about “sentient AI”; I’ve also spoken with CEOs that seemed to think AI could be panacea to all their technical problems. So there is some unnecessary froth around deep learning, and I hope these smaller bubbles will burst—the sooner the better—before they have time to grow bigger.

暂无回复
登录 或者 注册
相关帖子