回答问题人工智能源码_回答21个最受欢迎的人工智能问题

回答问题人工智能源码

Artificial intelligence sets the stage for a new era of solutions to be made with computers. It allows us to solve problems that we could not have imagined in the past. It’s a technology with massive potential and it’s also very confusing. To clear up some of the confusion I decided to make this video where I answer the most popular AI questions.

人工智能为计算机解决方案的新纪元奠定了基础。它使我们能够解决过去无法想象的问题。这是一项具有巨大潜力的技术，而且也非常令人困惑。为了消除一些困惑，我决定在这段视频中回答最流行的AI问题。

Who Invented Artificial Intelligence and where?

谁在哪里发明了人工智能？

The ideas about Artificial Intelligence evolved through centuries, starting with greek myths about Intelligent robots (Talos myth), but AI as we know it today only emerged in 1955. The term was coined by Alan Turing, Marvin Minsky, Allen Newell, Herber A.Simon, and John McCarthy.

关于人工智能的思想发展了数百年，从关于智能机器人的希腊神话(Talos神话)开始，但是今天我们所知的AI直到1955年才出现。这个名词是由Alan Turing，Marvin Minsky，Allen Newell，Herber A提出的。西蒙和约翰·麦卡锡。

Alan Turing, famous for his Turing Machine and working on deciphering the German Enigma Code.

艾伦·图灵(Alan Turing)以他的图灵机而闻名，他致力于解密德国《谜语》。

Marvin Minsky’s inventions include the first head-mounted graphical display and the confocal microscope.

Marvin Minsky的发明包括第一个头戴式图形显示器和共聚焦显微镜。

Allen Newell worked on two of the earliest AI programs, the Logic Theory Machine (1956) and the General Problem Solver

艾伦·纽厄尔(Allen Newell)从事两个最早的AI程序：逻辑理论机器(1956)和通用问题解决器

Herber A.Simon proposed a preferential attachment mechanism to explain power-law distributions.

赫伯·西蒙(Herber A.Simon)提出了一种优先依附机制来解释幂律分布。

John McCarthy created a programing language Lisp, invented garbage collection, and organized the Dartmouth conference where Artificial Intelligence was started as a field.

约翰·麦卡锡(John McCarthy)创建了一种编程语言Lisp，发明了垃圾收集，并组织了达特茅斯会议，在此会议上人工智能开始了。

2. How does artificial intelligence work?

2.人工智能如何工作？

Artificial intelligence is an all-encompassing term, which covers a myriad of different intelligent algorithms, of which the most popular is the neural network.

人工智能是一个无所不包的术语，涵盖了无数种不同的智能算法，其中最流行的是神经网络。

Neural Networks are built up by neurons. You can imagine them as a small computer chip that gets an input and based on some formula gives out some output.

神经网络是由神经元建立的。您可以将它们想象为一个小型计算机芯片，该芯片获得输入，并根据某些公式给出一些输出。

Let’s say for all numbers greater than 10 the neuron gives out 0 and for all numbers less than 10 it gives out 1. So, this could be an example of we have the data for rainy days and we are naively predicting forest fires. If there were less than 10 rainy days this year we predict a forest fire and give out the output of 1. If more than 10 then a forest fire is unlikely and the output is 0.

假设对于所有大于10的数字，神经元给出0，而对于小于10的所有数字给出1。因此，这可能是一个示例，即我们拥有雨天的数据，并且天真地预测着森林大火。如果今年的雨天少于10天，则我们将预测一场森林大火并给出1的输出。如果多于10天，那么一场森林大火的可能性就很小，输出为0。

Now, with more complex neural networks there are many many layers of such neurons that allow making extremely complex predictions. In that same example, we could have data for the humidity, temperature, amount of people that visited the forest, amount of thunderstorms, etc. All these elements would trigger different neurons and based on their outputs we could make better predictions.

现在，有了更复杂的神经网络，此类神经元的许多层都可以做出极其复杂的预测。在同一示例中，我们可以获取湿度，温度，参观森林的人数，雷暴的数量等数据。所有这些元素都将触发不同的神经元，并根据它们的输出可以做出更好的预测。

As a more concrete example of a neural network application, let’s take an image, which is a collection of pixel information that can be given as input into a massive neural network.

作为神经网络应用程序的一个更具体的示例，让我们来拍摄一张图像，该图像是像素信息的集合，这些像素信息可以作为输入输入到大型神经网络中。

Let’s say it was an image of a cat. The neural network is trained to look for different characteristics if it finds a cluster of pixels that represent ears it can already give out a prediction that this is some sort of animal. If it recognizes eyes, paws, etc. Other different clusters light up and give the prediction that an image of a cat has been given as an input.

假设这是一只猫的形象。如果发现发现代表耳朵的像素簇，该神经网络将经过训练以寻找不同的特征，它已经可以给出这种动物的预测。如果它识别出眼睛，爪子等，则其他不同的群集会亮起，并做出预测，表示已将猫的图像作为输入。

The learning of a neural network works by feeding a lot of inputs into the neural net and giving the output. By going through multiple iterations the model constantly adjusts it’s neurons, if we find ears in the picture we can be this much confident this is a cat. If it finds, ears, paws, nose, if it finds edges, shapes, the right colors, that increases the quality of the prediction.

神经网络的学习通过将大量输入馈入神经网络并给出输出来进行。通过多次迭代，模型会不断调整其神经元，如果我们在图片中找到耳朵，我们可以非常有信心这是一只猫。如果找到，则耳朵，爪子，鼻子，如果找到边缘，形状，正确的颜色，则可以提高预测的质量。

Basically, if we had that single neuron that gives a value based on a single data of information, in huge neural networks millions of neurons are given inputs, they are also adjusted based on training and outputs are being generated that can be generalized into a prediction. This is the basic gist of how neural networks work.

基本上，如果我们有单个神经元根据单个信息数据给出值，那么在巨大的神经网络中，将为数百万个神经元提供输入，它们也会根据训练进行调整，并且会生成输出，可以将其推广为预测。这是神经网络如何工作的基本要点。

3. Are artificial intelligence and machine learning the same?

3.人工智能和机器学习是否相同？

These terms can be easily confused as a lot of the time it seems they are being used interchangeably, but in short Artificial Intelligence is the broad all-encompassing term for different algorithms that do all kinds of tasks. While machine learning is a more narrow term, which focuses on training different models with data to do predictions. For example, face recognition.

这些术语在很长时间以来似乎可以互换使用时很容易混淆，但是总而言之，人工智能是用于执行各种任务的不同算法的广泛的无所不包的术语。机器学习是一个狭义的术语，它侧重于用数据训练不同的模型以进行预测。例如，人脸识别。

4. Why artificial intelligence is important?

4.为什么人工智能很重要？

When programming computers we usually have a very limited set of tools, if statements, if something is true do execute this branch, if not execute another branch. For loops, to execute a branch for a number of times and all the math operations.

在对计算机进行编程时，我们通常只有非常有限的一组工具，如果语句，如果为真，则执行此分支，如果不执行另一个分支，则执行该分支。对于循环，要执行多次分支和所有数学运算。

But when given a task to recognize an image we can’t really write a simple algorithm that would work. We can’t have an if statement for each pixel. So, for these kinds of tasks, we need sophisticated artificial intelligence algorithms. All in all, AI is bringing computer science to a new level where we are able to solve new kinds of problems that traditionally would be unsolvable.

但是，当赋予识别图像的任务时，我们实际上无法编写出可行的简单算法。我们不能为每个像素都有一个if语句。因此，对于这些任务，我们需要复杂的人工智能算法。总而言之，人工智能将计算机科学提升到了一个新的高度，我们能够解决传统上无法解决的新型问题。

All that allows computers to automate more and more tasks and expand their capabilities. Still, there is a ceiling at which AI can’t progress further, but I believe AI is going to set the stage for general AI as it’s going to be a whole new paradigm of a new era of problem-solving.

所有这些使计算机能够自动执行越来越多的任务并扩展其功能。仍然存在AI不能进一步发展的最高限度，但我相信AI将为通用AI奠定基础，因为它将成为解决问题新时代的全新范式。

5. Will artificial intelligence take over jobs?

5.人工智能会取代工作吗？

Yes, it will, just imagine all the drivers replaced with self-driving cars and trucks. It will take a while, but it will happen. And there’s going to be a bunch of fields that AI will disrupt, but this has happened throughout history time after time. Is it going to be different this time and it will be notoriously difficult to transition to new jobs? Time will tell.

是的，可以想象所有驾驶员都被自动驾驶汽车和卡车所取代。这将需要一段时间，但是会发生。 AI将会破坏很多领域，但这在历史上一次又一次地发生。这次是否会有所不同，并且很难过渡到新工作？时间会证明一切。

6. Will Artificial Intelligence replace doctors?

6.人工智能会取代医生吗？

Well, this is quite unlikely at least for several more decades. AI might help doctors do their job, giving care is also a very compassionate job. But until they will be fully replaced if at all is going to take a long long time.

好吧，这至少在几十年内都是不太可能的。人工智能可以帮助医生做好工作，护理也是非常富有同情心的工作。但是直到将它们完全替换，如果要花很长时间的话。

7. Will Artificial Intelligence replace Radiologists?

7.人工智能会取代放射科医生吗？

Radiologists are medical doctors that specialize in diagnosing and treating injuries and diseases using medical imaging procedures such as X-rays, MRIs, ultrasound.

放射科医生是擅长使用医学成像程序(例如X射线，MRI，超声波)诊断和治疗伤害和疾病的医生。

This is a field where AI can shine as it is exceptionally great at finding patterns. Now, it’s still far from replacing doctors, but it’s definitely a field that will get automated first.

这是AI能够发光的领域，因为它非常擅长于发现模式。现在，离取代医生还很遥远，但这绝对是首先要自动化的领域。

8. What programming language is used to code artificial intelligence?

8.使用哪种编程语言编写人工智能代码？

Artificial intelligence can be programmed in any language, but the community has formed mainly around Python, which has the cutting edge libraries for artificial intelligence like TensorFlow, Keras, Teanos, PyTorch, Numpy, Pandas, etc. and all the courses that you will find online will use Python. So, if you’re interested in AI learn to code with Python.

可以使用任何语言对人工智能进行编程，但是社区主要围绕Python形成，Python具有诸如TensorFlow，Keras，Teanos，PyTorch，Numpy，Pandas等人工智能的前沿库，以及您将找到的所有课程在线将使用Python。因此，如果您对AI感兴趣，请学习使用Python进行编码。

9. How artificial intelligence will change the world?

9.人工智能将如何改变世界？

It is already transforming the world, we are experiencing its effects every time we shop on Amazon and get recommended products to buy when we get Recommendations on YouTube when we talk with our phones. Or when the Government in China uses face recognition to track people. These unpleasant ways of how AI jumps into our lives will emerge even more in the future.

它已经在改变着世界，每当我们在亚马逊上购物时，我们都在经历着它的影响，当我们与手机交谈时，当我们在YouTube上获得推荐时，就会获得购买推荐产品的机会。或者当中国政府使用面部识别来追踪人时。 AI如何跳入我们生活的这些令人不快的方式将在未来出现。

10. How much do artificial intelligence programmers, engineers, and scientists make?

10.人工智能程序员，工程师和科学家能赚多少钱？

The AI field is booming and there’s a huge demand for talent. Salaries for general programmers are already high, but if you are also an AI programmer you can expect to earn from $150 000 per year to $300 000 and more. There have been reports of companies paying up to a $1 000 000 a year in salary for an AI scientist. Obviously, this is atypical, but if you are good at what you do, then you can definitely expect a very high salary.

人工智能领域正在蓬勃发展，并且对人才的需求巨大。普通程序员的薪水已经很高了，但是如果您也是AI程序员，则可以期望从每年15万美元增加到30万美元甚至更多。有报道称，公司每年为AI科学家支付的薪水高达100万美元。显然，这是非典型的，但是如果您擅长于自己的工作，那么您绝对可以期望获得很高的薪水。

11. What artificial intelligence can do?

11.人工智能可以做什么？

Artificial intelligence can recognize objects, your face, what you are saying. It can create pictures, sounds, music, text, it can even hold a conversation. It can do miraculous things, but even though it seems crazy, still we are far away from combining all these intelligent tasks into a thing that could do all of them and adapt to changing environments.

人工智能可以识别物体，您的脸部以及您在说什么。它可以创建图片，声音，音乐，文本，甚至可以进行对话。它可以做一些神奇的事情，但是即使看上去很疯狂，我们仍然距离将所有这些智能任务组合成可以完成所有任务并适应不断变化的环境的事情相去甚远。

12. Where artificial intelligence is used?

12.在哪里使用人工智能？

AI is used throughout many different fields. Chatbot and virtual assistant tech are solely based on AI language algorithms. Autonomous flying, self-driving cars, and autonomous vehicles are all powered by AI. Medical image analysis is a field of healthcare where AI is used. Warehousing, logistics, inventory management are all fields where AI is utilized. Shopping, social media, retail, fashion uses AI. Security and surveillance. Manufacturing and production. All these fields utilize AI.

AI遍及许多不同领域。聊天机器人和虚拟助手技术完全基于AI语言算法。自动驾驶，自动驾驶汽车和自动驾驶汽车均由AI驱动。医学图像分析是使用AI的医疗领域。仓储，物流，库存管理都是使用AI的所有领域。购物，社交媒体，零售，时尚使用AI。安全和监视。制造和生产。所有这些领域都利用AI。

13. Will artificial intelligence surpass human intelligence?

13.人工智能会超越人类的智力吗？

AI is really good when it is given a task and it has clear cut rules that can be evaluated as success or a failure. For example, chess. AI can easily learn the moves and start training as winning a match means it has done good and should do more of what worked and if it losses it should do less of that. Then train for millions and millions of times and you have the best chessplayer where no human can compete.

赋予任务AI并具有明确的规则，可以将其评估为成功或失败，这是非常好的。例如国际象棋。人工智能可以轻松地学习动作并开始训练，因为赢得比赛意味着它做得很好，应该做更多的工作，如果输了，应该做的更少。然后进行数百万次的训练，您将拥有最好的棋手，而任何人都无法竞争。

But what if there’s no way to know how to win and there’s no clear cut task at hand, then you can’t really train the AI. And that’s the main idea for calling it Artificial. Though the goal is to create general AI. An algorithm, that can quickly learn to recognize images and just as well be used to understand speech, then to talk and create music or art and find the answers to physics and math problems. Once we have that it will be a new era for humanity and I think we will be able to clearly see that our intelligence will be surpassed.

但是，如果没有办法知道如何获胜并且手头没有明确的任务，那您将无法真正训练AI。这就是将其称为人工的主要思想。尽管目标是创建通用AI。该算法不仅可以快速学习识别图像，还可以用来理解语音，然后进行交谈和创作音乐或艺术，并找到物理和数学问题的答案。一旦有了它，这将是人类的一个新时代，我认为我们将能够清楚地看到我们的智慧将被超越。

14. What artificial intelligence can’t do?

14.人工智能不能做什么？

Well, the main thing that AI can’t do is to adapt to different tasks, if it learns to drive a car the same AI won’t be creating deep fakes. As it requires a different kind of training. But in the future, it will merge up into general AI.

好吧，AI不能做的主要事情就是适应不同的任务，如果它学会驾驶汽车，那么相同的AI不会制造出严重的假货。因为它需要另一种培训。但是在将来，它将合并为通用AI。

15. Why artificial intelligence can be dangerous?

15.为什么人工智能会带来危险？

It’s very speculative and you can think of different scenarios some more likely than others, the first instinct is to think about general AI escaping into the internet, building itself a robot army and taking over the world.

它具有很大的投机性，您可以想到比其他情况更可能出现的不同场景，第一个直觉是考虑将通用AI逃逸到互联网中，自己组建一支机器人大军并占领整个世界。

It’s fun to make movies or write books about such scenarios, but honestly, I doubt such a scenario could unfold.

制作电影或写有关这种情况的书很有趣，但是老实说，我怀疑这种情况是否会发生。

What concerns me right now is how AI is used in surveillance of people in China, where cameras can identify you and track whenever you are doing something wrong. Their whole social credit system is very alarming and is a threat to human rights right now.

现在让我担心的是，如何将AI用于监视中国的人，在那里相机可以识别您并在您做错事时进行跟踪。他们的整个社会信用体系非常令人震惊，目前正在威胁人权。

IBM reported that they are shutting off their facial recognition services as they don’t align with their values.

IBM报告称，由于他们的价值观与他们的价值观不符，他们正在关闭面部识别服务。

https://techcrunch.com/2020/06/08/ibm-ends-all-facial-recognition-work-as-ceo-calls-out-bias-and-inequality/

In my previous video, I talked about deep fakes and how AI can recreate the voice of a person, how it can put your face into video footage and though I haven’t heard any such attack that has caused damage, it doesn’t take much imagination how this could work out.

在之前的视频中，我谈到了深层的伪造品，以及AI如何重现人的声音，如何将您的脸庞插入视频画面中，尽管我还没有听说过任何会造成破坏的攻击，但并没有想像力如何解决。

As AI evolves and we figure out more fields of application, more threats will arise, that aren’t as sexy as killer robots yet very disruptive.

随着AI的发展以及我们发现更多的应用领域，将会出现更多的威胁，这些威胁不像杀手级机器人那么诱人，但破坏力却很大。

16. How many companies use artificial intelligence?

16.有多少公司使用人工智能？

According to a survey in 2018 61% of businesses have implemented AI technology in their businesses. When I’m recording this video it’s already 2020, what a year and the numbers should be higher. I can only guess how much, but the thing is some businesses might not even recognize that they are using AI. If they have an online shop and a chatbot helping with customer support, well that’s AI right there, even though it’s not too obvious and doesn’t require an AI engineer to implement it. So, maybe the numbers are higher than reported in surveys.

根据2018年的一项调查，61％的企业已在其企业中实施了AI技术。当我录制此视频时，已经是2020年了，那一年和数字应该更高。我只能猜测多少，但问题是有些企业甚至可能没有意识到他们正在使用AI。如果他们有一个在线商店和一个聊天机器人来提供客户支持，那么那就是AI，即使它不是很明显，也不需要AI工程师来实现。因此，该数字可能高于调查报告的数字。

17. Can artificial intelligence be creative?

17.人工智能可以创造力吗？

Well, it depends on how you define creativity. If you would define it as being able to do things outside the realms it was trained to operate, then most likely no. But if you would define it as thinking outside the box and finding all the possible ways of winning then it’s definitely creative. To me, the best illustration of creativity is this video by 2-minute papers where AI was playing against AI a virtual game of hide and seek, and honestly, the results are quite amazing as they discovered plenty of outside the box ways how to win the game. For example, fly off the map. Closeout themselves. Block out the seekers and so forth. Is this creative well I will leave it to you to decide.

好吧，这取决于您如何定义创造力。如果您将其定义为能够在受过训练的操作范围之外做事，那么很可能不会。但是，如果您将其定义为跳出框框思考并找到所有可能的获胜方式，那么它绝对是有创意的。对我来说，创造力的最佳例证是这段2分钟的录像，其中AI与AI进行了一场捉迷藏的虚拟游戏，说实话，结果令人惊讶，因为他们发现了很多创新的方法游戏。例如，飞离地图。自己封闭。封锁搜索者，依此类推。这个创意好吗，我将由您决定。

18. How often do we use artificial intelligence?

18.我们多久使用一次人工智能？

If you are using your phone and computer daily, then you are most likely using some sort of artificial intelligence based technology. If you aren’t using voice recognition, then you might be using your camera, that uses AI filters, if you don’t do that then you might be doing google searches or shopping on Amazon or a ton of other things that use AI.

如果您每天都在使用电话和计算机，那么您很有可能会使用某种基于人工智能的技术。如果您不使用语音识别，那么您可能正在使用使用AI滤镜的相机，如果您不这样做，则可能是在Google搜索或在亚马逊上购物或使用AI的其他许多事情。

19. What artificial intelligence companies to invest in?

19.要投资哪些人工智能公司？

If you think artificial intelligence will change the world, then naturally you want to be apart of it and invest.

如果您认为人工智能将改变世界，那么自然就希望与世界分开并进行投资。

Though pure AI companies, that only work on AI research and products are quite rare.

尽管是纯粹的AI公司，但仅从事AI研究和产品研究的公司很少。

But pretty much all the tech giants are using AI to improve their products. Google, Microsoft, Amazon, Netflix, Apple, IBM, also companies like AMD and Nvidia are building the hardware that powers artificial intelligence.

但是几乎所有的科技巨头都在使用AI来改善他们的产品。谷歌，微软，亚马逊，Netflix，苹果，IBM以及AMD和Nvidia等公司都在构建支持人工智能的硬件。

And there are also startups that you could find to invest in, but that’s very risky.

还有一些初创公司，您可以找到投资的机会，但这风险很大。

20. How artificial intelligence is changing drug discovery?

20.人工智能如何改变药物发现？

Drug discovery involves processing a lot of data, doing research, analyzing previous papers, etc. Coincidentally AI is extremely good at analyzing data. And researchers are using AI to speed up the process and find new links between data that can lead to new discoveries.

药物发现涉及处理大量数据，进行研究，分析先前的论文等。巧合的是，AI非常擅长分析数据。研究人员正在使用AI来加速这一过程，并发现数据之间的新链接，这些链接可以带来新发现。

21. Where Can You Learn Artificial Intelligence?

21.在哪里可以学习人工智能？

Universities have the best cutting edge courses on artificial intelligence. So, you either should take a university course or look for some free university material online.

大学拥有最先进的人工智能课程。因此，您要么应该上大学课程，要么在网上寻找一些免费的大学材料。

Before you dive in you should be familiar with programming, in courses you are most likely to find code examples in Python.

在您开始学习之前，您应该熟悉编程，在课程中，您最有可能在Python中找到代码示例。

If you want to go deep and really understand the topic then Calculus will also be very important.

如果您想深入并真正理解该主题，那么微积分也将非常重要。

For example:

例如：

MIT course

麻省理工学院课程

http://cs231n.stanford.edu/

https://www.coursera.org/specializations/deep-learning/

You can also find a lot of random material here and there, but I would suggest to go through a more structured approach and only reach for different material if you want something to be explained in a different way to better understand it.

您还可以在各处找到很多随机材料，但是我建议您采用一种更结构化的方法，并且仅当您希望以不同的方式来解释某些东西以更好地理解它时，才可以使用不同的材料。

So, there you have it. The most popular AI questions answered. Though you still might have more. If so then leave them in the comments below and I will do my best to answer them, maybe I’ll even do a part two.

所以你有它。最受欢迎的AI问题得到了解答。虽然您可能还会有更多。如果是这样，请将它们留在下面的评论中，我会尽力回答他们，也许我还会做第二部分。