The “man-machine war” between AlphaGo and Lee Sedol quickly popularized the concept of artificial intelligence for the public.
But for Google, beyond playing Go, where is AI now? Will ARTIFICIAL intelligence defeat human beings in the future?
Greg Corrado, a senior research scientist at Google, recently traveled to China for interviews with media, including China Business News.
In his opinion, artificial intelligence and machine learning are both very important and fundamental new technologies at present. Many universities and companies around the world have their own ARTIFICIAL intelligence laboratories for continuous innovation and research and development, which benefit the whole industry. If the industry is dominated by only one company, it will grow more slowly and less efficiently, so more open competition in AI is a good thing, and Google wants the environment to remain open and competitive.
Giiso Information, founded in 2013, is a leading technology provider in the field of “artificial intelligence + information” in China, with top technologies in big data mining, intelligent semantics, knowledge mapping and other fields. At the same time, its research and development products include editing robots, writing robots and other artificial intelligence products! With its strong technical strength, the company has received angel round investment at the beginning of its establishment, and received pre-A round investment of $5 million from GSR Venture Capital in August 2015.
Greg Corrado told China Business News that there are some companies using ARTIFICIAL intelligence and machine learning for brand marketing. Some companies may go astray because they don’t have a clear understanding of what they are doing. But ultimately what consumers should care about is not how the technology is developed, but whether the technology actually works, and should not be easily swayed by marketing.
Talking about the challenges of ARTIFICIAL intelligence, he told reporters frankly, for the AI industry, the biggest difficulty lies in talent. “There are not enough engineers who can really understand and use the tools of ARTIFICIAL intelligence, and there are not enough innovative and business-savvy people. So we are more focused on how to cultivate and explore such talent.”
The following is a transcript of the interview:
Looking forward to new breakthroughs in machine learning
Q: When it comes to artificial intelligence, one of the hottest technologies is deep learning. How does machine learning help deep learning become more efficient?
A: The technology is constantly changing. Machine learning requires data samples, resources, tools, and computing power. Reviewing the development history of machine learning, it can be found that the efficiency of program operation cannot meet the demand due to the slow computing speed, high cost and technical reasons. Therefore, the development process of machine learning is also slow, and no actual products and services have been introduced. Until recent years, when computing power increased dramatically, speed increased, cost decreased, and applications became more widespread, that changed the picture. So now the bottleneck of machine learning is about people, people’s creativity and innovation ability, people who are good at and know how to use this technology. So our focus has changed, and we need to teach and train more people how to use machine learning to realize their innovative ideas, given all the other factors: sufficient data, free tools, resources, and sufficient computing power.
Q: At present, AI and machine learning are still limited to dealing with some relatively limited and specific areas of expertise. When do you think there will be more powerful general-purpose AI systems that can be applied to any area?
A: That’s an interesting question, and I think the future is going to be more specialized fields that use specialized techniques and models to solve specific problems and tasks. Such applications are more efficient and practical for a system and technology. I’m not particularly confident that a general-purpose technology will emerge, and even if it does, I don’t think it will be faster or more effective than a dedicated, targeted solution to a particular problem, but it will be slower and less efficient.
Q: What is the urgent breakthrough of deep learning?
A: Machine learning is not black magic, one of the most important is the need to make it easier for people to explore, allocate and different configuration variables (owing to the different needs to make a differentiation model adjustment), they don’t need to guess what is this black magic and working principle of behind, this will be the next development direction of deep learning, both theoretical research and engineering application upgrade, To better explore the concept and modeling of conjecture indicators in academic theoretical research.
Q: Which do you think is more effective in machine learning: manual supervised learning, unsupervised learning and semi-supervised learning?
A: At present, most or almost all of the machine learning systems that have been put into practical use are manual supervised machine learning (i.e. learning from samples collected by humans).
How to realize machine learning without manual supervision through automatic learning without specific samples is indeed very attractive in this research field. Many academic studies and papers related to it appear every year, but we have not found any achievements that have been put into practical application.
But I’d like to see innovative researchers figure out how to implement autonomous machine learning without human supervision in the future, and I can’t predict when. Of course, we really hope to see more talents engaged in this aspect of research, and also hope that there will be great progress and breakthroughs in the future.
Behind artificial Intelligence: Who are the biggest competitors?
Q: Google has recently launched a lot of new products about ARTIFICIAL intelligence. Does this signal that one of Google’s future priorities is to apply ARTIFICIAL intelligence to more products on a large scale?
A: I think needs to be stressed the point that artificial intelligence is not A separate independent existence, also is not Google one single machine learning system, but the past few years inside Google in all areas of product engineers adopted A kind of technical means, creative use of machine learning to optimize their research and development of products. It involves the application of new tools that these engineers learn how to use to optimize their systems for developing products. However, the specific methods and ways to use these tools will vary greatly from domain to domain due to the nature of the products and services.
Q: Previously, Google’s open source deep learning system TensorFlow, could you talk about Google’s future products and services on artificial intelligence and machine learning, and Google’s ideas?
A: For AI, I want to emphasize that it is not A specific product that can be packaged and sold. It’s actually a tool that software engineers and other creative people can use to create and develop new products and services. TensorFlow makes these basic tools that Google is using available to the public. For future products related to this field, Google intends to share its own platform with the public through cloud services. Through this cloud machine learning, other developers can develop and implement their own machine learning ideas, just like we do in Google. They can use our free software and tools through TensorFlow, or they can run their own machine learning systems using cloud services.
We will also provide pre-built machine learning subsystems to developers via the API, so that developers can implement techniques such as translation and image recognition with just a few lines of code. So developers don’t need to be machine learning experts to build their own machine learning applications.
Q: Nowadays, many companies advertise artificial intelligence when they launch their products and services, but the authenticity needs to be screened and considered. What is your opinion on this issue?
A: It is true that some companies are using ARTIFICIAL intelligence and machine learning for brand marketing, but ultimately what consumers should care about is not how the technology is developed, but whether the technology is actually working. If you really feel smart and useful by using something, don’t care how the technology works. So my suggestion is that consumers should be identified by the actual criteria of the product’s function, rather than being swayed by marketing, because it doesn’t matter.
Q: How do you see the artificial intelligence industry bubble?
A: As for the development of artificial intelligence and machine learning technology, I can deeply feel that they will continue to develop steadily and gradually in the following A long period of time, and will have A great impact on the technological innovation of various products and services. With the Internet, smart phones at the beginning of the development trend is very similar, and ultimately play an extremely important role. In the process, there may be some people who don’t know exactly what they’re doing and get sidestepped, which can easily happen in Internet companies or machine learning companies. However, the biggest dilemma facing the industry is not these, but there are not enough engineers who can truly understand and use artificial intelligence tools, as well as innovative and business-savvy talents, so we pay more attention to how to cultivate and explore such talents.
Q: Who do you think is Google’s biggest ai competitor right now?
A: The good news for all companies in the industry is that every company has A huge pool of talent. At the same time, many universities and companies around the world have their own AI LABS that are constantly trying to innovate and develop, so the whole industry benefits from that. If the industry is dominated by only one company, it will grow more slowly and inefficiently. So more and more open competition in ai is a good thing, and we hope it will continue to be open and competitive.
Can ARTIFICIAL Intelligence beat humans?
Q: Back in 2009, Google made an April Fool joke about Gmail’s ability to respond intelligently. Now, seven years later, the joke has come true. What kind of jokes do you think will come true in the next few years?
A: It seems I should really take A hard look back at some of the jokes we’ve been making over the years. The popularity of these ideas suggests to me that it’s not just engineers who need to learn how to use machine learning. I think anyone with an innovative mind, whether you’re in product development or a CEO in business, is more likely to come up with some great ideas and even dreams. These ideas might not have been possible five years ago, but it’s possible that technology will be able to turn them into practical applications by next year. It’s hard for engineers to do that on their own, but by working with people who think and see more broadly, they can build great products.
Q: What fields do you think artificial intelligence will have the most application value in the future?
A: I know that some related to Internet technology has been in the exploration and development, because it happens to be the origin of the machine learning inadvertently development field, but I think there are many other more field contains A lot of opportunities, may be A manufacturing, energy, health care, etc., these areas are in urgent need of the development status, machine learning has not involved too much.
So the most important thing is for the experienced business minded people in these industries to learn more about and learn how to apply machine learning so that they can see business opportunities and more possibilities.
Q: Do you expect artificial intelligence and machine learning to be as smart as, or even better than, humans?
A: With regard to the future of machine-human interaction, our future direction is to use machine learning to develop tools and technologies to assist and enhance our own human capabilities.
The idea of augmenting human intelligence has been around since ancient times, much as the pen and paper did for us: to aid our memory and enable us to accomplish more. But calculators, computers, web search, machine translation, image recognition and the like are built on top of those technologies to develop tools that extend some of our capabilities.
Google sees such features as the future of artificial intelligence and machine learning. That would change a lot of things. For an example of intelligent office, as the office of the people in the office, he needs a typewriter to type, need artificial check for spelling errors, need some by passing paper written work handover (paper-based office), but now we are beginning to use computers at work, the function of automatic spelling check, and email, These technologies have really changed things. They’ve helped people focus more on the newer, more important, more creative parts of their jobs.
Q: Does Google have some guidelines to ensure that AI technology moves in the direction you mentioned?
A: That’s why Google took the lead in setting up an organization called Partnership on AI to Benefit People & Society, an independent nonprofit organization, Many other companies are involved in creating an open platform for discussions about how AI interacts with people, society, and the economy, promoting understanding, discussion, and even open debate about AI. It is better to put the challenges of these topics on the table and discuss them openly than for companies to study them in private.
Giiso information, founded in 2013, is the first domestic high-tech enterprise focusing on the research and development of intelligent information processing technology and the development and operation of core software for writing robots. At the beginning of its establishment, the company received angel round investment, and in August 2015, GSR Venture Capital received $5 million pre-A round of investment.
Q: What do you think of quantum computing?
A: This is A technology that is still in the research stage and has no practical application until very far in the future, if ever. I think it’s just an amazing physics project right now, and it’s going to be a long time before it even makes it to engineering devices. If someone were to develop and build a quantum computer in their lifetime, it would make computing much more efficient, but all I can say for now is good luck to those working in the field.