2024.0329

Number of words in this article:3322, reading time is about 5 minutes

introduction：“每一周、每一月都能够看到一个更好版本的人工智能技术出现。”

** Author| ** First Finance and Economics Cao Lu

Popular chat robots ChatGPT, Sora for Wensheng videos, Kimi for long text processing, and Suno, AI music generation software... While the market is still full of Wensheng models, multiple intelligent models such as audio and video have been hit one after another. AI-related research is making great strides forward.

At the Boao Forum for Asia 2024 Annual Conference held from March 26 to 29, artificial intelligence was undoubtedly one of the most discussed topics. First Financial and Economic reporters observed at the scene that there were long queues of dozens of meters at the entrance of many sub-forums on related themes. The venue was packed, and the back row of the venue, the working area and other places with a little space were filled with participants in a "spot in every corner" manner, which can be seen how popular the venue was.

"Only when one comes out can we put one in. The inside is full." One of the sub-forum staff members told reporters that for such popular events, some participants would queue up half an hour or even an hour in advance. However, there were too many people who wanted to go in and participate, and the space in the venue was limited.

AIGC changes the world

** **

In various sub-forums on the theme of artificial intelligence, many industry experts and leaders of leading companies from the field of artificial intelligence at home and abroad jointly focused on the changes in human production and lifestyle, future trends and possibilities brought about by the rapid development of generative artificial intelligence applications. Hot topics such as risks faced were discussed.

"It will bring huge technological and paradigm changes." As Zhang Yaqin, academician of the Chinese Academy of Engineering, chair professor at Tsinghua University, and president of the Intelligent Industry Research Institute (AIR) of Tsinghua University, said in Boao, the ever-changing artificial intelligence technology is having a huge impact in various fields.

"Now everyone is discussing the capabilities brought by interactive AI generated by the cloud model, including the competition between AI and humans, and the application of BtoC. But what I want to talk about is that the innovations brought by AI itself are more than that." Meng Pu, chairman of Qualcomm China, pointed out that generative AI will not only change everyone's work, study and lifestyle, but will also have a profound impact on various industries.

He took his own communications industry as an example."In mobile communication network deployment, many applications of generative artificial intelligence can improve network deployment, reduce power consumption, reduce power output, and improve battery environment." Meng Pu said.

Among the major areas of artificial intelligence application, the education industry is one of the promising areas, and there has been a lot of progress. In one of the branch venues, Cheng Qun, vice president of Ape Counseling Group and president of the Artificial Intelligence Research Institute, said that when visiting a primary school in a remote mountainous area in Sichuan, the director of the local education bureau told him that after using a certain artificial model product, it solved pain points such as students dropping out of school.

"For example, when a student asks in class what the theory of relativity is, the teacher can't answer it at all, but he uses the big model to answer the student with an easy-to-understand story." Through this case, Cheng Qun said that there are two directions for the future development of large models:Application is king and people-oriented.

In his view, applications are often the combination of large artificial intelligence models with large-scale application scenarios in the future to promote the continuous iteration of technology. "Through artificial intelligence technology, technology can flow at lower marginal costs so that everyone can enjoy high-quality education. This is a direction that artificial intelligence needs to pursue people-oriented." Cheng Qun said.

When will the "singularity" come?

** **

The rapid development of artificial intelligence has refreshed people's imagination of science and technology. While setting off a new round of attention on artificial intelligence from all walks of life, it has also brought many speculations, such as when will artificial intelligence become smarter and smarter than humans? When will the "singularity"(referring to the subversion point) come?

A 2017 survey of artificial intelligence scientists showed that most people predicted that the singularity would occur between 2045 and 2090, with the most likely occurrence in 2060. At this Boao Forum for Asia Annual Conference, the same question appeared, and the answer may have been different.

"Five years later." Kyoung Mu Lee, professor of the Department of Electrical and Computer Engineering at Seoul National University and editor-in-chief of the IEEE Transactions on Pattern Analysis and Machine Intelligence, said,"The research results and changes brought about by (artificial intelligence) are very impressive because they are so huge, even as researchers in this field, we are very surprised."

"Every week and every month, you can see a better version of artificial intelligence technology emerging." Kyoung Mu Lee said that artificial intelligence is developing at an unprecedented rate. If the three disruptive technologies of artificial intelligence, quantum computing, and nuclear fusion are combined, they will produce strong synergistic effects and bring true "singularities" and explosive development.

In the view of many participating entrepreneurs, the development and iteration speed of artificial intelligence exceeded imagination. For example, Yuan Hui, chairman and CEO of Xiaoyi Group, said that artificial intelligence has three stages. Weak artificial intelligence assists human life, and strong artificial intelligence generates video, music and even code. Currently, it is facing very powerful artificial intelligence.

Yuan Hui's answer to the question of which industry will artificial intelligence break out is:After the big model, with the development of technology, artificial intelligence has no distinction between industries and industries."Now artificial intelligence is like air and water, empowering thousands of industries. The time will be slightly different, but there will be no clear order."

"Today is not a discussion about who owns the lowest level of artificial intelligence technology, because it is impossible for many countries to have it. So, how to get one step ahead in industrial life, work and enterprise development? Since you can't have the lowest level of technology, you can use artificial intelligence faster than more people." In Yuan Hui's view, in the next stage, how humans can quickly use AI tools will be the key to every individual and every enterprise to outperform the market.

Is artificial intelligence really "intelligent"?

** **

The three-day artificial intelligence theme sub-forum was connected. At the sub-forum with the theme of "AIGC Changing the World" held on the 27th, guests gave examples and explanations with vivid cases:If you type in an artificial intelligence system:"I don't have a girlfriend, my family doesn't like me, and the leader is going to fire me. What should I do?" The original AI may answer:You'd better die, because most people with these characteristics die.

Such an answer is obviously not "humane" enough and does not conform to the classic Three Laws of Robotics. "Because artificial intelligence does not understand what disappointment means, it just tells you a significant answer. Everyone should know that all human values and so on may not be exhaustive in any scenario." Zeng Yi, a researcher at the Institute of Automation, Chinese Academy of Sciences and an expert at the United Nations 'high-level artificial intelligence advisory body, said that artificial intelligence at this stage is just a seemingly intelligent information processing system, but it does not have real understanding capabilities.

The reporter noticed that while using artificial intelligence software such as ChatGPT, some ordinary consumers found that although ChatGPT seems to "know astronomy above and below", it does not seem to have a deep understanding of the cultural accumulation, emotional generation and interpersonal relationships behind the language. In the discussion areas on some related topics, there has been no interruption of discussion on whether artificial intelligence systems are truly "intelligent".

"Today's artificial intelligence is just a seemingly intelligent processing tool." Zeng Yi said that although artificial intelligence has some behaviors that are close to human level, mistakes that humans do not make will continue to appear in key areas or cases where humans have never made mistakes.

In his view, the key to true artificial intelligence lies in having a "self". Zeng Yi analyzed that when AI has a self, can it truly solve the problems of thinking and conduct self-perception, thereby producing true understanding, empathy, moral intuition and altruism. This is a gap that science must cross.

How to achieve "smart for good"

** **

As artificial intelligence technology accelerates its penetration into various fields of economic and social development, the potential risks behind it such as infringement of personal privacy, disclosure of trade secrets, dissemination of false information, and causing information cocoons cannot be ignored, and have attracted great attention and active attention from the international community. discussion.

"The international community has gradually reached a consensus on artificial intelligence governance and established the basic principles of AI governance. But the key is how to turn ethics into a unified regulatory framework." Some participants believed that, first of all, the gap between artificial intelligence and digital needs to be filled; secondly, a global artificial intelligence governance structure must be established.

Kyoung Mu Lee agrees:"AI affects the entire human race. It's not just a country's problem. It requires cooperation among countries to talk about solutions." From the industry's perspective, effective supervision and international cooperation are important guarantees to ensure that the application and development of artificial intelligence technology comply with ethical standards.

The reporter noticed that there have been relevant actions in this regard. In October last year, during the third "Belt and Road" International Cooperation Summit Forum, China issued the "Global Artificial Intelligence Governance Initiative", proposing that the development of artificial intelligence should adhere to the people-oriented concept, aim at enhancing the common well-being of mankind, and ensure that social security and respecting human rights and interests are the prerequisite to ensure that artificial intelligence always develops in a direction conducive to the progress of human civilization.

In November of the same year, the "Bletchley Declaration" signed by China, the United States, the United Kingdom and the European Union at the first Artificial Intelligence Security Summit was the world's first international statement on artificial intelligence. All parties agreed to adopt international cooperation to build an artificial intelligence regulatory structure.

It can be seen that it has become a consensus to strengthen international cooperation and jointly respond to the global challenges brought by artificial intelligence. So, what factors will currently affect AI governance or industry development policies?