The thing about machine learning is that bad data hurts a lot more than good data.

机器学习的问题是坏数据比好数据的伤害性更大。

Everyone, the left, the right, advocates, agencies agreed that change was being slowed down by bad data.

大家——无论是民主党、共和党、维权人士或有关机构——都觉得改变的步伐由坏数据拖累。

And everybody knows what bad data does.

每个人都知道不良数据的作用。

And at that point, we get bad data.

那时， 我们得到了错误的数据。

Bad data costs you time, it costs you money and it costs you resources.

不良数据会浪费您的时间、金钱和资源。

So bad data's basically how you get machine learning algorithms with unwanted bias.

因此，糟糕的数据基本上就是你获得带有不必要偏差的机器学习算法的方式。

But that's just like a problem with bad data, right?

- 但这就像坏数据的问题一样，对吧？

One comes from bad actors, so that's people using machines in evil ways, and the other comes from bad data.

一种来自不良行为者，即人们以邪恶的方式使用机器，另一种来自不良数据。

To avoid falling for bad data you need to make sure that the data source exists and that it's also credible.

为了避免陷入不良数据，您需要确保数据源存在并且可信。

Let's say we can solve issues of bad data and have facial recognition systems that work really well.

假设我们可以解决不良数据问题并拥有运行良好的面部识别系统。

So Sabrina, your issue is that you've built a tool that isn't working properly because it's been trained on bad data.

所以，Sabrina，你的问题是你构建的工具无法正常工作，因为它接受了错误数据的训练。

And in the rest of our logical lives, we would never make decisions on bad or highly abstract data, but we'll do that with feelings.

在我们余下的逻辑生活中，我们永远不会根据糟糕或高度抽象的数据做出决定，但我们会凭感觉去做。

Their official blog states that this was to prevent dislike harassment, but if you look at it through the lens of the algorithm, it's important to discourage users from providing bad data.

他们的官方博客表示， 这是为了防止不喜欢的骚扰，但如果你从算法的角度来看，阻止用户提供不良数据很重要。

And as the investigation unfolded, it became very clear that system had broken down and that...that system that we trusted to, you know, track down any flaw or any piece of bad test data really had failed.

随着调查的展开，很明显，这个系统已经崩溃了，我们所信任的可以追踪任何缺陷或任何不良测试数据的系统确实失败了。

I think that a wider swath of medical professionals have learned how to quickly interpret data and understand what is good data and bad data in a way that we have never been able to do before.

我认为更多的医学专业人士已经意识到了如何快速解释数据并理解什么是好数据和坏数据，以一种我们从未有过的方式。