Well, because we humans are the ones who produced the training data.

嗯，因为我们人类是产生训练数据的源头。

Details of the programming and training data of commercial ones like ChatGPT are proprietary.

像ChatGPT这种商业语言模型的编程和训练数据的细节为其公司所有。

That's what deep learning algorithms do as they go through all the training data.

这就是深度学习算法在处理所有训练数据时所做的事。

ChatGPT is designed to generate responses based on the input it receives and its training data.

ChatGPT 旨在根据其接收到的输入及其训练数据生成回复。

Once the training data had been cleaned up, the learning algorithms had to be tweaked as well.

一旦训练数据被清理干净，学习算法也必须进行调整。

They search through the training data to find related images and then copy over some of those pixels.

它们会通过搜索训练数据来找到相关的图像，然后复制其中的一些像素。

Wow, spooky. But that training data can also cause problems too, like by teaching the models to be biased.

哇，太诡异了。但是训练数据也会造成问题，比如教模型有偏见。

Or, in many AI systems, the robot learns from massive sets of training data.

或者，在许多人工智能系统中，机器人从大量的训练数据中学习。

Their training data uses art and books created by artistsand authors without their consent.

他们的训练数据未经他们的同意就使用了艺术家和作者创作的艺术作品和书籍。

It's  ignoring ChatGPT's training data limitations.

它忽略了 ChatGPT 的训练数据限制。

Green, 7 arms Training data can have correct labels on them.

绿色，7 条手臂训练数据可以有正确的标签。

The more training data it's given, the better the algorithm becomes at correctly identifying new images.

提供的训练数据越多，算法就越能正确识别新图像。

So how much does it matter who is choosing the training data to feed in the algorithms?

那么，谁选择训练数据来输入算法有多大关系呢？

That the hardest, most difficult questions, are ones that are surprising or very different to the training data.

最难、最困难的问题是那些与训练数据令人惊讶或非常不同的问题。

By identifying and condensing prevailing styles in their training data, generative AI models are playing a similar role.

通过识别和压缩训练数据中的流行风格，生成式 AI 模型正在发挥类似的作用。

So if the key to AI intelligence isn't just more training data, what is it?

那么，如果人工智能的关键不仅仅是更多的训练数据，那么它又是什么呢？

In order to train our machine learning classifier to make good predictions, we’re going to need training data.

为了训练我们的机器学习分类器做出好的预测，我们需要训练数据。

This will help extend those models to work for languages where there isn't any labeled training data available to train the system.

这将有助于扩展这些模型，使其适用于没有任何标记训练数据可用于训练系统的语言。

So, to understand the AI's test performance, we should know what's inside of the training data.

- 因此，要了解人工智能的测试性能，我们应该知道训练数据的内容。

There's always a line you can draw between its response and what was in its training data.

Musser：在它的响应和训练数据之间，你总是可以画一条线。