科学美国人60秒 这种人工智能可以像婴儿一样学习

（单词翻译）

This is Scientific American's 60-second Science, I'm Christopher Intagliata.

这里是科学美国人——60秒科学系列，我是克里斯托弗·因塔利亚塔。

Artificial intelligence systems have bested humans at chess, poker¹, Jeopardy², Go, and countless³ other games.

人工智能系统已经在国际象棋、扑克、危险游戏、围棋和无数游戏中击败了人类。

But machines still aren't that great at understanding some basic rules about the physical world.

但机器在理解物理世界的一些基本规则方面仍然不够好。

They still can't do what 3-month-olds do.

机器仍然做不到像3个月大的婴儿那样。

And I'm a champion for babies at the end of the day and this is a clear win for babies.

说到底，我只能当婴儿中的佼佼者，婴儿明显是胜利者。

Babies are still slam dunking even our most powerful computers when it comes to intuitive physics.

当涉及到直观物理学时，婴儿仍然会碾压我们最强大的计算机。

Cognitive⁴ psychologist Susan Hespos of Northwestern University listed off a few examples of those "intuitive physics" principles.

西北大学的认知心理学家苏珊·赫斯波斯列举了几个“直觉物理学”原理的例子。

Like "solidity" - your coffee cup does not just fall right through the table.

比如“固体性”，咖啡杯不会穿过桌子掉下去。

Or "continuity" -- objects don't just blink in and out of existence.

比如“连续性”——物体不会在眨眼之间消失。

And "boundedness" -- when you pick up your coffee cup, it sticks together.

比如“捆绑性”——当拿起咖啡杯时，会粘连在一起拿过来。

You don't end up with just the handle.

不会只拿过来杯子的把手。

Babies know all three of these things as early as three months of age.

婴儿早在三个月大的时候就知道这三件事。

Their visual acuity⁵ is lousy, the world is blurry--they could barely grasp this stuff.

他们的视力很差，世界很模糊，他们几乎不能理解这些东西。

You know, babies get a lot of things wrong.

你知道，婴儿会做很多错事。

But it's these initial kernels⁶ that get elaborated and refined through experience in the world.

这些最初的内核，通过世界上的经验得到了充实和完善。

Now computer engineers have taken a page from the baby playbook.

现在，计算机工程师们已经从婴儿的剧本中借鉴了一页。

Researchers at DeepMind -- the AI company that trained computers to beat humans at Go -- have endowed a machine learning system with certain kernels of knowledge about intuitive physics built in--akin to what an infant might be equipped with.

DeepMind是一家人工智能公司，该公司训练计算机在围棋中击败人类。这家公司的研究人员为机器学习系统内置了直觉物理学的某些核心知识--类似于婴儿可能具备的东西。

And after watching the equivalent of just 28 hours of training videos, showing things like balls rolling, and blocks dropping -- the AI system actually showed "surprise" when it was shown something physically⁷ impossible.

人工智能系统观看了相当于28个小时的训练视频，这个视频展示了球体滚动和砖块掉落，在此之后，当系统看到一些物理上不可能的事情时，它实际上会表现出“惊讶”。

Its counterparts not modeled on babies weren't as sharp.

其他不以婴儿为模型的人工智能系统就没那么灵敏了。

It's really interesting that when you do this direct comparison what you find is learning from experience goes far. But only so far.

很有趣的是，当你做这个直接的比较时，你会发现从经验中学习是很有意义的。但也仅仅到这里而已。

And the computer that was built based on research on babies, did far better.

基于对婴儿的研究而制造的计算机，表现要好得多。

It's confirming evidence for what baby research has shown for a while.

这证实了婴儿研究一段时间以来所显示的证据。

The results appear in the journal Nature Human Behavior.

该研究结果发表在《自然人类行为》期刊上。

Hespos wasn't involved in the work, but wrote an editorial accompanying the paper.

赫斯波斯没有参与这项工作，但为该论文撰写了一篇社论。

She says the research is a step towards making machine learning systems more efficient thinkers -- like humans. Even the tiny ones.

她表示，这项研究朝着让机器学习系统像人类一样更高效地思考迈出了一步。即使是很小的一步。

Thanks for listening for Scientific American's 60-second Science. I'm Christopher Intagliata.

谢谢大家收听科学美国人——60秒科学。我是克里斯托弗·因塔利亚塔。