In 1951, Alan Turing, the father of computer science and the father of artificial intelligence, had a conversation with a friend who did some simple machine learning things, but found it very difficult to teach machine learning, either doing it wrong, doing it slowly, or doing nothing. Alan Turing's friend asked him a question: Who is learning? Are machines learning, or are people learning?
At the 2023 Inclusion Bund Conference, He Jifeng, academician of the Chinese Academy of Sciences, told this story to the audience in his speech, and Turing felt that both humans and machines are learning, humans guide machines, make machine learning more intelligent, and in the guidance process, humans also slowly learn what kind of guidance enables machines to meet the goals of human needs. At the end of his speech, He Jifeng emphasized: "In the process of artificial intelligence development, we must not forget the core role of human beings."
The just-concluded Bund Conference attracted visitors from all over the world with a science and technology exhibition area of nearly 8,000 square meters, nearly 20 academicians of the "two academies", Nobel Prize and Turing Award winners, and 500 executives and experts and scholars of leading science and technology enterprises. Data security, privacy computing and other contents are often talked about, and how to solve the trust problem in the intelligent era has attracted attention from all parties. Generative AI and large models are still hot topics in this forum, and while looking forward to the arrival of the "new world" of science and technology, human beings have many questions that need to be thought about first.
At this high-profile FinTech Summit, the application of technology in the industry and the scenarios that may change have triggered countless discussions. What should we think about AI? How to learn to maintain the core role of humanity? What kind of "new world" is it moving towards?
Expectations "New World"
Many "black technologies" appeared at this Bund Conference.
The exhibition area of the future pension apartment shows the all-round intelligent equipment for the elderly, self-service medical cabin, AI brain training equipment, etc., and preliminarily shows the future technology development direction of detecting the physical condition of the elderly by pace and screening people for Alzheimer's disease risk by playing games. Unexpectedly, many young people became "heavy experience players" in this exhibition area, and they believed that this could really be their future retirement scene.
Wearing a special helmet, the system can recognize the brainwave waveform characteristics of the experiencer and generate its exclusive "brain pattern", which is a brain-computer head-mounted nuclear technology interactive technology jointly launched by Ant Security Lab, China Academy of Information and Communications Technology and brain-computer cooperative manufacturer Boruikang. In this technology exhibition area, the experiencer realizes the "brain pattern" generated by the idea to unlock the equipment, and the "brain pattern" will become their unique identity ID. "In addition to unlocking, is it possible to achieve mind control of objects through this technology in the future?" Some experienced people immediately asked.
Whether it is the general audience, industry participants, professional scholars, they are looking forward to the new technology for the future more changes will happen soon.
"A mobile phone is a bank branch." Wu Lianfeng, Vice President and Chief Analyst of IDC China, released the "Five Trends of Banking Digital Technology" at the Bund Conference: portable banking, AI risk control, digital employees, edge IoT and cloud native architecture.
According to the "2022 China Banking Service Report", the average electronic diversion rate of banks is 97%, and more than 90% of their business has been online. IDC predicts that by 2025, more than 80% of banks will deploy digital humans, providing 90% of customer service and financial advisory services. By 2026, applying artificial intelligence to credit will become the mainstream choice for banks. Due to the application of more complex models and algorithms, cases such as bank fraud and money laundering will decrease by 12% by the end of 2026.
Jing Xiandong, chairman and CEO of Ant Group, is concerned about the breakthrough of privacy computing. He revealed that Ant is focusing on privacy computing, hoping that one day in the future, the cost and efficiency of encrypted computing will be the same as today's plaintext computing, which means that the industry can realize the exchange of data value, rather than the data itself, "that era will really usher in the big explosion of the data age." ”
The future development of traditional container trade is also tied to the fast ship of blockchain development. Chen Sijia, CEO of the Global Shipping Business Network, said at the Bund Conference that the blockchain electronic bill of lading has moved from the PoC verification (proof of concept) stage to the window period of large-scale commercial application.
In February this year, the Global Digital Container Shipping Association announced that nine of the world's top 10 shipping companies have committed to digitize 50% of the original bills of lading within 5 years and 100% by 2030, thereby accelerating the digitalization of container trade.
Chen Sijia believes that by actively introducing blockchain technology and using the blockchain network to circulate electronic bills of lading, ensure the authenticity of the content and circulation of bills of lading, greatly simplify the operation process, and may shorten the time required for the release of goods by sea from the traditional 2-3 days to 1-2 hours.
"The essence of AI is collective"
This year's hottest big model in the science and technology industry is also expected by financial practitioners, Ant Group released its financial model at the Bund Conference, and explorers of financial big models such as Du Xiaoman, Hang Seng Electronics, Flush, and Haitong Securities also appeared at the Bund Conference.
Liu Shufeng, chairman of Hang Seng Electronics, said that the changes of large-model technology to wealth management services may be the whole business chain, and will play an important role in both the "investment" and "advisory" ends, as well as risk control, operation, quantitative trading and other fields. From the perspective of internal operations, the big model refreshes the new paradigm of software research and development, and in the long run, 80% of the code may be generated by artificial intelligence, and 20% of the core code at the architecture level will be provided manually.
"AI's potential benefits to the global economy will reach $25 trillion, which is one of the most important tracks for all businesses today, but this era has just begun." Daniel Ni, chairman of McKinsey China and global senior managing partner, predicted on the Bund that the technological storm of generative AI is expected to open a new round of technological and industrial changes related to the next 8-10 years.
However, while embracing the big model in the financial industry, which pays great attention to the rigor of domain knowledge and professional logic and has a very low tolerance for errors, there are also many voices at the Bund Conference that are cautious about the "illusion" of the big model and rethinking the relationship between humans and AI.
Wang Weiqiang, general manager of the machine intelligence department of Ant Group and chief scientist of Ant Security Lab, said that the problem of factual deviation "illusion" caused by the inability of large models to distinguish between true and false, ethical problems caused by data bias, copyright problems in the process of knowledge learning and generation, and the resulting internal security issues for enterprises are all great challenges for practitioners. At the same time as the development of large models, it is indispensable to detect the security of large models, the construction of trusted capabilities, and even the development of security defense technologies.
The heads of privacy protection of Baidu, Ant, Douyin and other companies also gathered at the Bund conference to discuss how to do a good job in privacy protection in the wave of new technologies such as big models.
Nie Zhengjun, chief privacy officer of Ant Group, said that in the face of the risks and challenges brought by large models, enterprises can use anonymization tools in the training stage of large models to prevent the risk of privacy leakage or misuse of personal information brought by future models themselves. It is also necessary to pay attention to separating the data generated by the service party and the data used in the self-operated business when the commercial application of the large model is implemented.
Xu Quanquan, head of data compliance at Baidu, suggested that the next step in privacy protection may be more about taking the road of industry collaborative governance. Every enterprise has an unshirkable responsibility to fulfill its responsibilities in privacy protection and data compliance, and at the same time, it must cooperate with upstream and downstream partners to participate in privacy protection and data compliance.
"The essence of AI is collective, not individual. It is just a new engineering system, science and technology is a very good new engineering system, is universal, and is fair, just, transparent, stable, these characteristics are the characteristics of the entire engineering system, not the characteristics of the individuals in the system. "Machine learning titans," Michael Jordan, academician of the National Academy of Sciences and the American Academy of Engineering, said.
"I don't think AI will replace people." Michael Jordan said. In his view, the real problem facing artificial intelligence is how to achieve collective intelligence in a collaborative, decentralized system, and how humans can design this collective intelligence to solve its uncertainty.
To some extent, this collective design and collaboration is the biggest issue facing human beings in maintaining their core role in the rapid development of science and technology. It is the part that humans also need to learn while the machine is learning.
Source of this article: Yicai
Scan the code to follow us