The hottest Justin Kassel, don't be afraid of AI.

Justin Kassel: don't be afraid of artificial intelligence, its future is in our hands

Justin Kassel: don't be afraid of artificial intelligence, its future is in our hands | 2018 easy future technology summit

17:53 source://

original title: Justin Kassel: don't be afraid of artificial intelligence, its future is in our hands | 2018 easy future technology summit

easy technology news on September 21, The "all in era" of the "2018 easy future technology summit" hosted by easy and easy technology was held in Beijing today. Justine Cassell, the queen of artificial intelligence and deputy dean of the school of computer science at Carnegie Mellon University, made the future opportunities and surprises of artificial intelligence

Justin Kassel asked: artificial intelligence brings so much fear to human beings, where do these fears come from? Will these fears come true? What is the actual situation

she believes that robots will mechanize our work, and robots will indeed replace part of human work. But at the same time, we should also see that AI will also bring us some new jobs: jobs that need innovation, jobs that need social skills and collaborative jobs will increase

"our living standards have been greatly improved, personal health has been improved, urban congestion and pollution have been alleviated, workers' productivity has been improved, and the efficiency of government operation has been improved. But this can only be achieved if everyone in the city benefits from artificial intelligence." Justin Kassel said

what is artificial intelligence? Justin Kassel believes that AI is not a software, but a method; In the process of building machines, machines can do things like people or do things that people can do. AI tools include planning, reasoning, symbolic rules, machine learning and big data

Justin Kassel believes that AI + people will be the future. This future will bring new production mode, new education mode, new work and communication mode, new travel mode and intelligent infrastructure

therefore, Justin Kassel believes that through artificial intelligence, human beings can obtain new methods, new tools and new powerful computing power to realize increasingly complex computing; And the data that is constantly generated all the time will also be obtained by the computer

as humans spend more and more time on electronic devices, science and technology need to return to the old but still important principle: manage long-term interaction with people in the way of establishing relationships with people, and establish a growing relationship. Justin Kassel believes that this problem is very thorny. Deep learning alone is not enough. We need to observe the behavior of real people to build a socially conscious computer

finally, Justin Kassel believes that artificial intelligence is in the hands of human beings

the following is the full text of the speech

Justin Cassell:

thank you, good morning, everyone

I'm here today to tell you about the future of artificial intelligence, both in terms of opportunities and the surprises it will bring to mankind. Now people have a lot of fear about AI and robots. You can see from this screenshot that this is a screenshot of many newspaper headlines. People are very afraid of and reduce accidents. What problems AI brings, can you think of the opportunities it brings, think about whether it will be completely destroyed, and whether robots will replace us. This fear has become so great that such images have been filled in it, Maybe you also have such a fear. These robots that originally serve you are lurking in your office. Open the window and replace you. Where does this fear come from? And will this fear come true? Will robots really replace us

in fact, if you observe the current economic situation, you will see a completely different reality. Indeed, robots will make our work more automated, just as factories, including printers and electricity, bring us automation. Electricity actually has some negative effects on workers, because workers work longer hours, and robots will certainly have a little negative impact on our work

what you see here is the latest research done by the economist. To study the impact of robots on our working environment, in fact, some jobs will disappear. These jobs are relatively repetitive, cyclical, and do not require high skills and boring work. But we seldom hear the other side is that some new jobs will be created, and these jobs need very human skills, that is, the ability to cooperate with others. The ability of teamwork and social skills. Let's look at this curve. The work of this curve represents some non repetitive work, which requires creativity, and the most technical jobs that require social skills will have the largest growth

I was talking with a Chinese Minister in charge of economy just now. We talked with him about our hobbies and discussed the latest developments in China. Chatting like this is actually very important in our work. Sometimes it's also a luxury, a kind of gossip. The reason why we are human is that we have social needs. If we can have this kind of emotional connection with others, we can better cooperate and communicate. Therefore, employment in this area will continue to improve as AI replaces some relatively simple jobs

in fact, AI will bring a lot of progress, including economic progress and progress in our lives. There are also some congestion and pollution in our city, including helping us save energy. Most of these changes are positive. But we should also remember that not all the changes brought about by these changes are positive, but I also want to emphasize one point, that is, the second point I want to say. The first point is that social ability, human characteristics and human excellences will always be important. The second point is that all these will not force us, not that we have no choice, this is a necessary fate. Instead, we have choices, and we are proud of our choices, which will force us to rethink everything and the meaning of work. Do we work 24 hours a day? For example, India has such a system, or does France start to implement such a system as soon as we hear that the working time is only 5 hours, or do we work 2 hours a day and then hand over the work to robots. If robots can do everything we do, what is the meaning of human beings and the definition of human beings? In an AI era. With the dramatic changes in AI life in the future, our definition of human beings and our understanding of all kinds of things will change

what is artificial intelligence? Artificial intelligence is not software. In fact, we often talk about prejudice when we do artificial intelligence. For example, my headphones and microphone are not designed for women, and women with long hair are not easy to wear. What is artificial intelligence? There were many discussions at the Davos Economic Forum last year. I organized a round table discussion on machine learning. At that time, the guests included kings, CEOs and presidents of various countries. At that time, I asked him if you could give me a definition of machine learning? No one can answer it. Finally, one person said it was very intelligent. I can give such an answer at most 1 point. If it is 10 points, full score. What is artificial intelligence and machine learning? Artificial intelligence is not a certain software. It is a method. It is a way to think about what computers can do. In the 1950s, when artificial intelligence was just born, these computer scientists, psychologists, anthropologists and mathematicians came together to think about whether it can be a kind of software, and can imitate human behavior, Or it can accomplish what human beings can do

artificial intelligence is a method rather than a special tool, and machine learning is the strongest tool in AI. So what is machine learning? Now we have a higher understanding of AI than those famous CEOs, kings and presidents, because we already know the answer. There is little difference between artificial intelligence and statistics and mathematics. Everyone learned these in high school, or what is the difference between them and the accounting tools we use? When you predict the income of your company in one year, five years later and ten years later? Statistics is a formula. You enter numbers into the formula, and then you get a value. In machine learning, you build a lot of software. These software search for enough examples, and you help them give hundreds of forces. After reading them, they will automatically give an answer, and then they will be able to find similar examples by themselves. As machine learning, this machine can find numbers that need to be filled in formulas, not only numbers, but also pictures, behaviors, words But bick has its own design advantages. What belongs to data can be put into machine learning software, and machine learning can automatically program for itself. This sounds a little scary, but for me, most of it saves time

so this algorithm learns from data. This opportunity allows us to understand the world in a way that we humans can't do. Maybe I can see more than 100 numbers, and then come up with a formula. Through the formula, I know the meaning of these 100 numbers, but I can't process onemillion numbers and onemillion photos at the same time. I can't take into account everyone here, talk to each of you or find the similarities and common ground of each of you. But machine learning can be expanded. This is an opportunity

but what are its negative or potential risks? If these figures, photos and data we put into the machine are not enough, there is no way to represent all of us. For example, if I only give photos of its CEO in the algorithm, most CEOs are male, so when it learns the definition of CEO in the process of learning the algorithm, it will recognize that men are CEOs. This is certainly not the prejudice we want to see, but the prejudice that exists in the current world. Through this algorithm, it will enter the future world

we can give you an example of Google face recognition software. We can see whether a photo can recognize a face or not, and can distinguish whether it is a face. We can give it 1000 or 5000 examples, but most of the examples are white, so this algorithm makes a decision because the CPU of the computer that white people are human beings, and all human beings are white. Then we showed the software a black CEO, who was identified as a monkey. This is very bad, which is called algorithm bias. When we give AI algorithm data, we must be very careful to ensure that it can represent all of us. It should be sufficiently representative

in addition, our speech recognition and speech recognition also depend on AI