杰弗里·辛顿

“We should recognise that this stuff is an existential threat. And we have to face the possibility that unless we do something, it might wipe us out.”

我们应该认识到这东西是生存威胁。我们必须面对这种可能性：除非我们采取行动，否则它可能消灭我们。

“In the meantime, I'd say it's going to be a long time before it's as good at physical manipulation as us and so a good bet would be to be a plumber.”

与此同时，我认为AI在物理操作能力上赶上我们还需要很长时间，所以一个不错的选择是去当水管工。

“I don't believe we're going to slow it down. And the reason I don't believe we're going to slow down is because this competition machine... is making you go faster and faster.”

我不相信我们会放慢速度。我不相信会放慢的原因是，这台竞争机器……正在让我们越跑越快。

“What we really need is a kind of world government that works run by intelligent, thoughtful people. And that's not what we got.”

我们真正需要的是一个由明智、深思熟虑的人领导的、运作良好的世界性政府。而我们并没有这样的政府。

“If it can do all mundane human intellectual labour then what new jobs is it going to create? You'd have to be very skilled to have a job that it couldn't just do.”

如果它能完成所有普通的人类智力劳动，那它会创造出什么新工作呢？你必须非常有技能，才能拥有一个它无法直接胜任的工作。

“It's very hard to get your head round the fact that we're at this very, very special point in history. We're in for a relatively short time, everything might totally change.”

很难让你真正理解，我们正处在一个历史上非常、非常特殊的时刻。在相对较短的时间里，一切都可能彻底改变。

“If you can get that gradient information, we know that we can then train a big system that starts with random weights to do wonderful things. The brain needs to get information like that, and it probably gets it in a different way from the standard algorithm used in these AI models which is called back propagation.”

如果你能得到那个梯度信息，我们知道我们就能训练一个从随机权重开始的大系统去做奇妙的事情。大脑需要得到那样的信息，而且它获取信息的方式可能不同于这些AI模型中使用的标准算法——反向传播。

“We like somebody who has this really cute tiger cub. It's just such a cute tiger cub. Now, unless you can be very sure that it's not going to want to kill you when it's grown up, you should worry.”

我们就像某人养了一只非常可爱的虎崽。它真是只可爱的虎崽。现在，除非你非常确定它长大后不会想杀死你，否则你应该感到担忧。

“I think if the public realized what was happening, they will put a lot of pressure on governments to insist that the AI companies develop this more safely.”

我认为，如果公众意识到正在发生的事情，他们将向政府施加巨大压力，要求AI公司更安全地开发技术。

“Previously what the large language models had to do is they spit out one word at a time and that would be it. Now, they spit out words and they're looking at the words they spit out... that's called chain of thought reasoning. And so now they can reflect on the words they spit out already and that gives them room to do some thinking in.”

以前大语言模型要做的就是一次吐出一个词，仅此而已。现在，它们吐出词语并观察自己吐出的词语……这被称为链式思考。因此现在它们可以反思自己已经吐出的词语，这给了它们一些思考的空间。

“I always believed that we had to understand how the brain worked, to understand how the mind works.”

我一直相信，我们必须理解大脑是如何工作的，才能理解心智是如何运作的。

“As long as you follow your curiosity, you're going to be fine.”

只要你追随自己的好奇心，你就会没事的。

“Original ideas come from noticing that people are doing things wrong or that you think people are doing things wrong.”

原创想法来自于你发现别人做错了事情，或者你认为别人做错了事情。

“I think if you have to figure it out, you're not really passionate about it.”

我认为，如果你需要去弄清楚自己是否热爱某事，那你可能并不是真的热爱它。

“We're not [rational beings]. We're great big analogy machines. We work by seeing analogies.”

我们并非（理性生物）。我们是庞大的类比机器。我们通过识别类比来运作。

“One of them was very interesting, and the other one worked.”

其中一个非常有趣，另一个则真正奏效。

“The point is to show that even in a tiny net that you can have two different minima.”

目的是表明，即使在一个微小的网络中，你也可以有两个不同的最小值。

“The way we're going to interpret a binary image is we're going to clamp the binary image on the visible units and then we're going to go round updating neurons using this probabilistic decision rule.”

我们解释二值图像的方法是：将图像固定在可见单元上，然后使用这个概率决策规则循环更新神经元。

“So the answer is everything it needs to know about the other weights is conveyed by letting the system settle to the equilibrium in these two phases.”

答案是，它需要了解的关于其他权重的所有信息，都是通过让系统在这两个阶段稳定到平衡状态来传递的。

“So you can think of it as like an enzyme that helped deep learning be born.”

所以你可以把它想象成一种酶，帮助了深度学习的诞生。

“Between five and 20 years from now, there's a good chance, a 50% chance we'll get AI smarter than us.”

未来5到20年，有50%的概率我们将获得比人类更智能的AI。

“Anybody who says it's all going to be fine is crazy, and anybody who says they're inevitably going to take over is crazy.”

说‘一切都会好’的人是疯子，说‘AI必然会接管一切’的人也是疯子。

“We're making things more intelligent than ourselves. We've never been in that situation before.”

我们正在制造比自己更智能的东西，这是人类从未面临过的处境。

“If you have an idea and it seems right to you and it's different from what everybody else believes, don't give up on it until you've figured out why it's wrong.”

如果你有一个想法，你觉得它是对的，但与所有人的信念都不同，那就坚持下去，直到你弄清楚它为什么是错的。

“It's ridiculous in this rich country like Canada that 20% of the indigenous communities do not have safe drinking water.”

在加拿大这样富裕的国家，20%的原住民社区没有安全饮用水，这太荒谬了。

“Once they realize getting more control is good and once they're smarter than us, we'll be more or less irrelevant.”

一旦它们意识到获得更多控制权是有益的，一旦它们比我们更聪明，我们就会变得无足轻重。

“Multimodal chatbots can already have subjective experiences.”

多模态聊天机器人已经能够拥有主观体验。

“We're not special. And we're not safe.”

我们不再特殊，也不再安全。

“Understanding is converting the words into feature vectors.”

理解，就是将词语转换为特征向量。