Is there a copyright?
How to effectively protect users' privacy information and data security?
More and more powerful artificial intelligence technology,
While bringing convenience and surprise to people,
It also raises corresponding risk issues and hidden dangers such as intellectual property rights, data security, and technical ethics.
On May 27, the forum "Intellectual Property Protection in the Digital Era - Starting from Intellectual Property Issues and Risks in the Use of ChatGPT" was held at the Shanghai Science Hall, hosted by the Shanghai Association for Science and Technology, the Shanghai Municipal Administration for Market Regulation and East China University of Political Science and Law, and organized by the Legal Advisory Committee of the Shanghai Association for Science and Technology, the Shanghai Science Education Exhibition Technology Center (Shanghai International Science and Technology Exchange Center) and the Shanghai Huangpu District Market Supervision Administration.
Ye Qing, President of East China University of Political Science and Law and Director of the Legal Advisory Committee of the Municipal Association for Science and Technology, presided over the opening ceremony, and Han Zhiqiang, Secretary of the Party Group and Vice Chairman of the Municipal Association for Science and Technology, and Peng Wenhao, Member of the Party Group and Deputy Director of the Municipal Market Supervision Bureau, delivered speeches. At the awarding ceremony, Peng Wenhao, member of the Party Group and Deputy Director of the Municipal Bureau of Market Regulation, awarded the Shanghai Science Education Exhibition Technology Center (Shanghai International Science and Technology Exchange Center), which won the first batch of "Shanghai Trade Secret Protection Demonstration Station".
Liu Feng, Director of the Intellectual Property Business Committee of the Municipal Lawyers Association and member of the Legal Advisory Committee of the Municipal Association for Science and Technology, presided over the forum activities, and four industry experts delivered keynote reports on ChatGPT and the intellectual property issues and risks in its use.
To engage in artificial intelligence, we must have "code farmer" thinking:
Tight timing logic + mechanization
Professor Wei Hui of the School of Computer Science and Technology of Fudan University made a keynote report entitled "Opportunities and Problems of Artificial Intelligence Entering the ChatGPT Moment", starting from the history of artificial intelligence, focusing on the analysis of the basic principles of ChatGPT and the opportunities and problems of generative artificial intelligence. Professor Wei Hui hopes that everyone will understand the characteristics of artificial intelligence and how computers learn. In his report titled "Opportunities and Problems for Artificial Intelligence Entering the ChatGPT Moment", he said that ChatGPT is a generative artificial intelligence that can generate new content, but basically cannot be called innovation, and it may be more appropriate to call it "brainstorming or side solicitation". It will bring many opportunities, such as lowering the threshold for some jobs, but also bringing a lot of problems, such as plagiarism or false information, and even creating public opinion. However, artificial intelligence is not so magical, it is just a more complex program, and the things that can be done are reasoning, problem solving, decision-making, understanding (studying images, studying language) and learning. To engage in artificial intelligence, we must have "code farmer" thinking, that is, "formalizable + algorithmic + algorithm can be calculated", and artificial intelligence is not "intelligent".
Computer learning is completely different from human learning. Human learning pays attention to one inference, through one example to transform into three other applications, artificial intelligence is the opposite, machine learning needs to give many examples, may train a thousand highly homogeneous examples just to solve a single model problem, and different application goals of the learning strategy, implementation details are not universal.
Professor Wei Hui said, looking at it this way, do you think it is "stupid". Things that people think are easy are huge challenges for artificial intelligence, such as dialogue, scene understanding, uncertain decision-making, etc., and doing a single task and doing a variety of tasks are completely different things. Don't forget that people have highly optimized brains that have undergone a long biological evolution, and computers are just a system. So it relies on a large amount of data to train, and the amount of computation is huge. The so-called large language model is to calculate the probability of a large number of words and word collocations distribution, and word collocations are almost endless, and the amount of things to be remembered by the computer is "too large".
On the other hand, why is it that artificial intelligence is easy to do what people find difficult, for example, why can computers learn Go on their own and show super ability? "Not because the computer is smart, but because the rules of Go are very simple, the game of Go is very simple to determine the victory or defeat, the data is very standardized, the algorithm is easy to accurately define, win to get a high score, lose to get a low score, and the algorithm pre-defines what kind of follow-up operations correspond to wins and losses, this is reinforcement learning, because winning and losing are easy to judge by themselves, so the machine can score itself." Professor Wei Hui said.
Let's discuss the copyright risks of generative AI
Focusing on the topic of intellectual property issues and risks in the use of ChatGPT, the participating experts expressed their opinions. Cong Lixian, Dean of the Intellectual Property School of East China University of Political Science and Law and member of the Legal Advisory Committee of the Shanghai Association for Science and Technology, delivered a report on the theme of "Copyright Risk and Governance of Generative AI", analyzing the copyright risk of generative AI from three stages: data mining, content generation and content completion, and put forward his own opinions on the copyright risk governance of generative AI-generated content.
Cong Lixian said that artificial intelligence should still be in the "infant stage", not the stage of "giant" super strength. But even "infants and young children" may pose copyright risks. Copyright protection has two risks, on the one hand, the risk of infringing the copyright of others when it generates content, and on the other hand, the risk of copyright infringement of the copyright of the work it generates.
Xu Chunming, Professor of Shanghai International Intellectual Property Institute of Tongji University and Director of the Intellectual Property Discipline Committee of Tongji University, delivered a keynote report entitled "Intellectual Property Legal Issues and Reflections on Artificial Intelligence", discussing the relationship between the development of new technologies and intellectual property law, the impact of artificial intelligence on intellectual property objects, and the impact of artificial intelligence on intellectual property subjects. "We often see results without focusing on what the process of generation looks like," he says. After listening to Professor Wei Hui's explanation, I have a deeper understanding of artificial intelligence. Generative AI, the process of generating things, is more based on a probability of calculation, rather than expression. ”
"The laws of countries around the world stipulate that copyright works must be the creation of natural persons and human beings, and are the expression of the author's thoughts and feelings. The invention must be the invention and design of natural persons and human beings, and the intellectual achievement of the inventor. For example, he said, the American inventor Taylor has applied to many patent offices around the world, and he has used DUBUS as the inventor, and has been rejected in the United States, the United Kingdom, Germany, Japan, and the EPO on the grounds that the inventor is not a natural person under patent law.
Qian Guangwen, member of the Trial Committee of the Shanghai Intellectual Property Court and president of the Second Comprehensive Trial Division of Intellectual Property Rights, made a keynote report entitled "Judicial Protection of Data Intellectual Property Rights in the Era of Big Data", sharing the nature of big data, the ownership of property rights and interests in big data, and the judicial protection mode of big data intellectual property rights.
In the roundtable discussion and on-site interactive session, they discussed how to promote knowledge sharing while protecting intellectual property rights in the digital era, discussed and expressed their views on the fair use of ChatGPT, data collection and privacy, international intellectual property issues, whether the content generated by ChatGPT is a work, copyright ownership, how to protect users' personal privacy information and trade secrets, and how to protect data security and privacy. Wang Qian, professor of East China University of Political Science and Law and vice chairman of the Academic Committee, participated in the event as a dialogue guest. Wang Qian stressed that the Copyright Law is used to encourage people's creation, and the content generated by artificial intelligence is not a work.
How to promote the development of artificial intelligence industry
At the event, a number of experts called for appropriate adjustment of relevant laws to adapt to the new changes brought about by the development of artificial intelligence and promote the healthy development of artificial intelligence in China.
In response to some people's proposals, in order to promote the development of the artificial intelligence industry, the copyright law should be used to protect the generation of artificial intelligence. Wang Qian said that the United States does not recognize AI-generated content as works, which does not dampen the enthusiasm of American companies to invest in AI. Think about it, how can Open AI and Microsoft make money with ChatGPT? The money earned by users in paying is small money. Microsoft has embedded ChatGPT into various products, and Office is charged, and more people are willing to pay for Office. Microsoft will also license other businesses to get involved in ChatGPT extensively to provide smart services to their users, which is a lot more money.
Wang Qian said that training ChatGPT will definitely feed a lot of data, which is basically equivalent to putting the entire Internet content into it. Like Microsoft, Open AI, training artificial intelligence, throwing so many works into it, obviously not private copying. Google used as a union member to train artificial intelligence, whether to obtain the permission of the writer, this matter of the two sides of the point of view, the writers' union said that it must be licensed, this contradiction has not been resolved.
"European and American copyright owners can rely on collective management organizations and collective bargaining to fight for the interests of the latter, which is not feasible in China, and it is difficult to solve the problem through collective management organizations or collective bargaining. For China to develop artificial intelligence, it is more appropriate for the Copyright Law to recognize that this act of copying works solely for training is fair use, which is more realistic. ”
As one of the important activities of the "National Science and Technology Workers Day" of Shanghai Association for Science and Technology in 2023, this forum aims to explore the new path of intellectual property protection in the digital era under the background of unprecedented challenges in the creation, management and use of massive intellectual property (data) in the digital era, and to explore a new path of intellectual property protection in the digital era around "big data, artificial intelligence and intellectual property", promote enterprise scientific and technological innovation, and serve scientific and technological workers' innovation and entrepreneurship.
Material provided: Shanghai Science Education Exhibition Technology Center (Shanghai International Science and Technology Exchange Center)
Review: Personnel Department of Municipal Association for Science and Technology (Veteran Cadre Work Department)