Dialogue Singularity University President: Within 15 years, artificial intelligence has reached human intelligence.

"Robots can reach human levels in the 2029."

Zhang Duo (hereinafter referred to as Zhang): At this year's CES show, we saw a drone that could take off from the palm of the hand, the resolution of the TV was 8K, like at CES last year or 4K, and the number of robots on display at CES this year, though many, But none of them is stunning. The robots at the show were still cold, and the robots of Star Wars "Sripio" and "Olympus" were slow to come. So in the current sense of the robot as the main direction of the process of artificial intelligence is slow, now the robot inventors have encountered what technical bottlenecks?

Rey Kuzwer (hereinafter referred to as Ray): Each time we create a new artificial intelligence technology, it is no longer called artificial intelligence, but called speech recognition, robotics, image recognition, or character recognition. Artificial intelligence has not yet come to human levels, but it has been able to do something remarkable.

A good example is IBM's supercomputer "Watson" was able to take part in an American quiz show called Dangerous Edge, a very esoteric language game, in which Watson's knowledge comes from 20 billion language documents, including Wikipedia, in the human-computer war, There is a rhyme that asks for answers from the question: the bubbling pastry ingredient delivers a long, tedious speech. Watson answered, "Meringue-harangue (the chatter of meringue)"

It's the right answer, but this is a difficult question, the world's most powerful two human players did not answer, Watson's final score than those two "the strongest brain" score combined more, the machine after the explosion of mankind.

Google unmanned cars have been safely driven for 1 million miles without incident. It used to be thought that only humans could drive. It would be easier if we let Watson's class of robots drive our cars.

Now computers are doing a good job of identifying, analyzing, and making accurate text descriptions of complex pictures, and they can do it, but computers are doing well. Ai has progressed much better today than it did five years ago.

Robots can now do complex things, naturally avoiding obstacles and passing through a messy room. I always say that the robot can reach the human level in 2029, this is not 2029, we wait for another 15 years.

They have a lot of things we humans can't do, can you find out what you want from 2 billion of web documents minutes and seconds?

They can translate all kinds of languages, although not as good as our human translation, but not bad.

I often use machine translation to read emails written in other languages, and their translation results can be understood. They are also slowly progressing.

If artificial intelligence does not do what human beings can do, humans will be very mad: this machine is so stupid, not even that!

A few years ago, machines had been able to tell the difference between cats and dogs.

So every time something goes wrong with the computer, we throw it aside and complain, "It's not smart enough." So you see the news reports of artificial intelligence and people in the daily life of real artificial intelligence is still a gap, so the AI in everyone's eyes looks very tall, but the real AI really should do so tall, Perhaps its biggest use today is just Google or Baidu.

Explosive material: Artificial intelligence has been used in the research of military weapons

Zhang: How should human beings cope with the crisis caused by the high development of artificial intelligence? Whether it can only be like "Terminator" and other movies show that the robot to torture human beings. Will future highly intelligent robots and humans be able to get along well?

Ray: At the moment humans and machines are still in harmony, but I think the AI has been used for military weapons research. But you see those about the future of science fiction, the routine is nothing more than AI appear, and then fight with humans, or two humans to fight for an AI and kill, those are cheat children, AI robot and human is not one or two people, in fact, 2 billion people.

People who have smartphones now know everything about the world through simple operations, a skill we didn't have a few years ago. Today, an African child with a smartphone knows more than 15 years ago in China or the U.S. president, so these technologies are powerful.

In fact, I think, Ai makes the world more peaceful. There is a book called "The Decoupled Angels of our Nature", the author of Steven Pinker in that book is about the exponential decline in violence. Your chances of dying in a war or being hacked to death today are much less than they were a few centuries ago. Part of the reason is that our communication with people is more intelligent now, so our understanding of others is more abundant, the contradictions are much reduced and the world is more harmonious.

How far are we from being immortal?

Zhang: The future with a variety of implant technology, artificial and virtual body technology mature, human life expectancy will also be extended, so how far away from the immortality of our generation still hope to keep up?

Ray: For the immortality of aging and disease, I would like to say three points:

First, there are many things we can do to keep young and overcome disease. For my part, I was diagnosed with type II diabetes 30 years ago, but I developed my own disease-resistance program, which has not been a disease for 25 years.

The second aspect is genetic technology, the redesign of our life program, fundamentally our life is software, our genes are data sequences, like software design.

There are 230,000 software programs called "genes" in our human body, which we used to have a lot of time with. Now we can program these "software". We can turn off the harmful, pathogenic genes by RNA interference. We can use gene therapy to add new genes or change old genes so that we can evolve very quickly.

Gene therapy is also a kind of information technology in a way, its development follows my proposed "the acceleration cycle rule", is expanding rapidly outward at exponential speed, the computer performance will ascend one times each year, the biotechnology actually is same. It's been more than 10 years since we started mapping the genome, and some of the results have been used in clinical practice. Health and medicine will become very different after 10-15, when we can turn off genes and genetic programs that cause disease and aging.

Across biology, the third area we talk about nanotechnology. Millions of nano-robots in our bloodstream defend our immune systems like T-cells, but they're smarter, for example, the immune system doesn't fight cancer, and it feels like it's your normal cell so it won't kill them. Our immune system is also ineffective against retrovirus, which is explained in many different ways. But we can create an intelligent nano robot that is smarter than T cells so that it can shut down the aging gene to accomplish the tasks that T cells can't do.

Aging is not inevitable, it is a process we can control, but not 1 processes, there are 11 kinds of human aging process, for each process we have to deal with the strategy, now there is a strategy to slow them down.

We sometimes see some seventy or eighty-year-olds, I'm 68 years old now, but my biological tests are younger than my physical age, and we also see older people who look older than they are.

With today's knowledge can slow down the aging, but in the future, we will have more powerful technology to really shut down from the source of aging let us forever young, eventually we will have virtual body to face some virtual environment. So we can not only determine the length of life, but also determine the breadth of life. Machines and people will blend into one, and we will live in augmented reality with a virtual body like avatar. By then, people hundreds of of miles away could sit next to us with the help of those virtual indistinguishable.

The future computer will enter the body and brain, the brain will connect with the cloud

Zhang: The current stage of the machine is becoming more and more intelligent, in the future they may also gain the consciousness of autonomy, no longer listen to people. But as an artificial intelligence optimist, I think that 2029 years from now, humans are going to go beyond those highly intelligent robots, and I know you're very optimistic about that. Can we predict how long it will take for humans to control the runaway agents after the singularity?

Ray: I think the computer 2029 can reach our human intelligence, in fact, I always really think. Like everyone in today's AI circles, the prediction I made 1999 years ago, at a meeting in Stanford, when other AI scientists reached a consensus that was hundreds of years old, my 30-year vision was a standout.

Today, I can guess that many people's ideas are 20 or 30 years later, but with the current pace of technological development, 15 years is a lot.

But in the 30 's, we will be smarter than the smart ones, even though most of the computers are not in our bodies or our brains. But almost anyone is doing it, and some Parkinson's patients are implanted with a chip (brain pacemaker) in their brains to restart their brain, which is suppressed by Parkinson's disease.

But most computers are not in our bodies at the moment, but they are close to our bodies. I was on the belt (showing the cell phone on his belt), and I could take it out and approach our human technology, which made me smarter than I was 10 years ago, eventually getting into our bodies and our brains, By using them we can also connect the brain to the cloud, and by extending our personal thoughts to the cloud, we think about the problem more deeply. That was 2030. The realization of the global brain will make us humans smarter. Although most computers are still in our body, this process is already happening.

20: Humans Conquer Cancer

Zhang: Please predict the next year in which humans will be able to conquer cancer?

Ray: I have a project with MIT scientists for 8 years, and we've identified the cause of cancer, a special type of cell--cancer stem cells that breed other cancer cells that can cause a tumor. Today's therapies include chemotherapy and radiation therapy. These treatments do not kill cancer stem cells, which cause cancer, as you destroy all the bees in a hive except the Queen bee, but the Queen bee also breeds more worker bees and fills the hive again. This is why cancer can recur, which is why cancer spreads because of this particular cancer stem cell.

After we identified the causes, we let them grow in petri dishes, and we've identified a lot of drugs that can kill them, but we have to learn to master their kind and proportions like cocktails. We will soon have our next experiment at the medical school in Wisconsin. This is a project we have done on cancer.

No matter what I'm doing, don't let it get you down, there are a lot of exciting projects like this in the world. We have found the most basic cause of cancer, and the medical technology to conquer it will then be exponentially developed. On the basis of information processing. We do everything we can to control cancer.

Cancer is a hereditary disease, not just a naturally acquired gene at birth, but also a gene that passes through your newly proliferating cells throughout your life. Adult cells are not the same as the children's cells, your cells are proliferating, they multiply, sometimes mistakes, some areas of the error will trigger those stem cells to produce cancer cells have many plans to control the problem.

We should treat it as a manageable disease, such as a chronic illness, and not a fatal illness.

In 10 years, I'm sure we can really conquer cancer in 20 years. So, children, you have a chance to catch up.

(Note: This interview was originally issued on January 22, the love of Qi Yi technology channel)

(Responsible editor: Mengyishan)