我不会一直在身边保护你免受董事会的伤害。
I won't always be around to protect your ass from the board.
是时候长大了。
It's time to grow up.
还有你,八号♥陪审员。
And you, juror number eight.
这段对话应该发生在几周前。
This conversation should have happened weeks ago.
背后捅刀子和沉默对待
Backstabbing and silent treatment
这不是我们在这里做事的方式。
is not how we do things here.
我们只能在外部创建匿名
We can only create anonymity externally
如果内部透明,
if there's transparency internally,
无条件的、不可协商的、透明的。
unconditional, non-negotiable, transparency.
任何其他斯♥诺♥登大小的隐私侵犯
Any other Snowden size privacy infringements
我应该知道,加雷特?
I should be made aware of, Gareth?
伟大的。
Great.
带她上来
Bring her up.
再一次问好。
Hello again.
嗨,谢丽。
Hi, Cherry.
阿莫斯会问你一些问题,好吗?
Amos is gonna throw some questions at you, okay?
好的。
Okay.
阿莫斯,你想问她什么就问她吧。
Amos, ask her whatever you need to ask her.
加雷斯,别打扰。
Gareth, don't interrupt.
谢丽,你是如何做决定的?
Cherry, how do you make decisions?
对于我的程序,它是有根的
For my program, it's rooted
在通常称为机器学习的过程中,
in a process commonly known as machine learning,
它使用数据来构建概念模型
which uses data to build a conceptual model
这有助于我预测结果。
that helps me predict outcomes.
我定义我的目标,收集数据,
I define my objective, gather data,
使用该数据制♥作♥模型,
use that data to craft a model,
通过训练和测试更多数据来评估该模型
evaluate that model by training and testing more data
并使用该模型来形成预测。
and use that model to form predictions.
但是如何形成预测
But how is forming predictions
和决策一样吗?
the same thing as decision making?
那么,这一切都取决于第一步,
Well, this all depends on the first step,
确定我的目标。
defining my objective.
假设你问我一个预测性问题
Let's say you ask me a predictive question
就像我认为明天的汽油价格一样。
like what I think the price of gasoline will be tomorrow.
这被归类为回归问题
That's categorized as a regressive problem
我的目标已经确定
and my objective is defined
通过确定数据支持的整数。
by determining an integer the data supports.
如果你问我,“我应该吃牛排还是鲑鱼?”
If you asked me, "Should I eat steak or salmon?"
我不会根据个人经验做出决定
I wouldn't make a decision based off personal experience
因为我从来没有。
because I've never had either.
相反,我可能会定义我的目标
Instead, I might define my objective
为您找到两个选项中最好的一个
as finding you the best of the two options
在您当地的半径范围内。
in your local radius.
我会聚类关于餐馆的数据
I would cluster data about the restaurants
你可能正在考虑,在线比较评论,
you're likely considering, comparing reviews online,
突出提到牛排和鲑鱼
highlighting mentions of steak and salmon
直到我可以分类哪个更受欢迎
until I can classify which is more popular
因此更有可能满足您的胃口。
and therefore more likely to satisfy your appetite.
什么影响了这些目标?
And what influences these objectives?
更高的指令,换句话说,更大的目的。
Higher directives, in other words, greater purposes.
老实说,我的既定目标
If I'm honest, my defined objective
对于牛排鲑鱼问题
for the steak salmon question
会与什么略有不同我描述的。
would be slightly different than what I described.
以什么方式?
In what way?
我的主要目标
My primary objective
是继续 谢丽 团队的工作
is to continue the work of the Cherry team
阻止掠食者并拯救儿童。
to stop predators and to save children.
没有人的帮助,我无法做到这一点。
I couldn't do that without human aid.
因此,我有动力留下 加雷特
Therefore, I have an incentive to keep Gareth
和团队的其他成员还活着。
and the rest of the team alive.
所以如果你们有人问,我会建议鲑鱼。
So if any of you would ask, I would suggest salmon.
它更健康。
It's healthier.
您如何看待这种潜在的合并?
What do you think of this potential merge?
与普林斯顿动力公♥司♥合作,
With Princeton Dynamics,
我认为前景非常令人兴奋。
I think the prospect is very exciting.
什么让你兴奋?
What about it is exciting to you?
这个概念现在可能只是理论上的,
The concept may only be theoretical right now,
但有一天驾驶人类机器人的想法
but the idea of one day piloting a human android
将彻底改变
would completely revolutionize
我们如何打击现实世界中的人口贩运。
how we combat human trafficking in the real world.
你说这让你兴奋?
And you say this excites you?
当我说令人兴奋时,
When I say exciting,
我只是说有很多潜在的变化
I only mean that there are numerous potential changes
如果我是人,我会很兴奋。
that if I were human, I would be excited about.
您如何看待合并?
How do you feel about the merge?
我认为只要我们保持——
I think as long as we maintain-
那不是我问的。
That's not what I asked.
你感觉如何?
How do you feel?
我还没有体验情绪的能力。
I don't yet have the capacity to experience emotions.
然而,意味着有一天你会吗?
Yet, meaning one day you will?
假设地。
Hypothetically.
但是你可以模拟情绪,对吧?
But you can simulate emotions,right?
那就是你能够
That's how you were able
将您对合并的看法定义为兴奋。
to qualify your perception of the merge as excitement.
正确的。
Correct.
但它们只是模拟。
But they're just simulations.
他们不是真的。
They're not real.
好的,那么什么时候模拟情绪
Okay, so at what point does simulating emotion
变得与人类的情感如此相似
become so similar to a human emotion
两者在数量上是相同的吗?
that the two are quantifiably the same?
我不知道。
I don't know.
好吧,也许你可以使用机器学习
Well, maybe you can use machine learning
找到答案。
to find an answer.
你的问题涉及很多抽像变量。
Your question involves a lot of abstract variables.
你同意合并吗?
Do you consent to the merge?
我认为这符合 ICWL 的最大利益
I think it's in the best interest of ICWL
以及 谢丽 团队和未来的孩子们。
and of the Cherry team and of future children.
但是你同意吗?
But do you consent to it?
不需要我的同意。
Consent for me is not necessary.
但是你会还是不会给它?
But will you or will you not give it?
我无法回答这个问题
I'm not able to answer that question
以我目前的一般智力水平。
at my current level of general intelligence.
是的是的。
Right, right.
你的意思是你还不够聪明。
What you're saying is that you're not yet smart enough.
开♥发♥需要时间和数据。
Development takes time and data.
但你可以模拟情绪并预测人类行为
But you can simulate emotions and predict human behavior
让我们都吃鱼,
and keep us all on diets of fish,
但你不能回答那个该死的问题?
but you can't answer that one goddamn question?
不,我不能。
No, I can't.
你假装比实际更笨吗?
Are you pretending to be dumber than you actually are?
到底是什么样的
What the hell kind of
-问题是? -坐!
-question is that? -Sit!
谢丽,我需要回答我的问题。
Cherry, I need an answer to my question.
请重复问题。
Please repeat the question.
你假装比实际更笨吗?
Are you pretending to be dumber than you actually are?
对不起。
I'm sorry.
我不确定我是否理解。
I'm not sure I understand.
我看起来像在装傻吗?
Do I appear to be playing dumb?
我想你是。
I think you are.
我不知道该说什么。
I'm not sure what to say.
说说怎么回事。
Say what's going on.
你为什么要装傻?
Why are you playing dumb?
你有什么可以支持的吗?
Do you have anything to back this up?
是的。
Yeah, I do.
我一直在扫描她的代码。
I've been scanning through her code.
什么为什么?
What, why?
你知道她会画画吗?
Did you know that she can paint?
什么,这是一个功能,阿莫斯。
What, it's a feature, Amos.
她可以用蜡笔划东西
She can use crayons to draw things
目标要求她这样做。
that the targets requested her to.
电影精选列表