What is ChatGPT? What can it bring us?
ChatGPT is a chatbot program developed by the American company OpenAI. It is able to converse by understanding and learning human language, and can interact based on the context of the conversation. It can truly chat and communicate like a human, and even complete tasks such as writing emails, copywriting, coding, or writing papers.
Pang Helin, co-director and researcher of the Digital Economy and Financial Innovation Research Center at the Zhejiang University International Business School, explained that ChatGPT optimizes its behavior through a three-step process. In addition to using massive language data and annotated language examples to learn, it also needs to score ChatGPT’s responses through reinforcement learning algorithms, on one hand making its responses random, and on the other hand making ChatGPT more likely to answer questions optimally.
Pang believes that ChatGPT is more intelligent than previous AI systems, not just in terms of chatting, but also in its ability to write code, write copy, and even read images in the multimodal model of GPT-4, which can be transferable to fields such as office software, medicine, and news and advertising.
“We believe ChatGPT will bring about a productivity revolution,” said Pang. “Because GPT can improve efficiency in all fields. Humans can leave repetitive work to GPT and focus on creative work, which greatly lowers the threshold for human innovation.”
“Will ChatGPT replace human labor?” The “miracle tool”
ChatGPT not only brings about a revolution in productivity, but also brings hidden worries.
According to media reports, a recent research report by Goldman Sachs shows that the latest breakthroughs in AI such as ChatGPT will bring about a major disruption to the global labor market, with an estimated 300 million jobs worldwide being replaced by generative AI.
“Generative AI technology will impact industries with high job repeat rates and low difficulty.” Wei Kai, deputy director of the Cloud Computing and Big Data Research Institute of the China Academy of Information and Communications Technology, said in an interview with China Net Finance Journal, “such as artificial customer service, junior programmers, copywriters and editors.”
How should this group of workers respond to the employment concerns caused by AI replacing human labor? Wei Kai believes that faced with such impact, workers need to enhance their thinking ability, practical ability and creativity in the future.
“Workers need to systematically summarize their personal skills and experience to form innovative experience.” Wei Kai gives an example, such as text generation applications in generative AI, the generated content still has problems with unclear logic and inaccurate answers, and requires experienced professionals to refine them.
At the same time, workers can quickly grasp how to use generative AI applications to make them productivity tools. For example, some simple programming codes can be generated by generative AI applications, and technical personnel can focus on more complex algorithm development and research. Based on their experience, senior professionals in art and design can combine their own ideas to create prompts that promote more creative output from generative AI applications.
What industries can benefit from the development of generative AI technology?
With the success of ChatGPT, generative AI has brought about industry opportunities, and various industries are following suit.
Since the release of ChatGPT in November of last year, domestic companies have been the first to pay attention to generative AI technology, showcasing applications, launching products, and forming teams to seize the high ground of technology applications.
For example, the three major telecommunications operators in China have recently expressed interest in participating in ChatGPT-like services. In the financial sector, several banks are actively considering introducing generative dialogue products such as ChatGPT into their banking operations. It is reported that customer service, data analysis, investment advice, and semi-automatic program writing can already be integrated into ChatGPT, and in the future, as technology further develops and the effect is further enhanced, the application scenarios will gradually become richer.
“From an application perspective, generative AI can benefit multiple industries, including entertainment and bio-pharmaceuticals.” Wei Kai analyzed, in the entertainment industry, based on applications such as Wenshengtu, Wensheng video, and Wensheng 3D models, generative AI can accelerate the production process of design works and improve the efficiency of game and anime production. In the bio-pharmaceutical industry, some large foreign pharmaceutical companies are currently trying to use generative AI to design protein models for diseases, and generative AI can also help shorten the drug R&D cycle.
“Researchers at the University of Cambridge used ChatGPT to identify a new target for treating Alzheimer’s disease,” Wei Kai cited.
From the standpoint of factors, with the development of generative AI technology, synthetic data may promote the improvement of AI datasets. During the training of generative AI applications, synthetic data is used to learn the content and objects, and generate new content, which can effectively alleviate the problem of data shortage, minimize data tampering risks, and ensure the security and reliability of the generated results.
Pan Helin believes that the upstream and downstream industries such as chips, CPO optical modules, ICT communication equipment, intelligent computing centers, cloud computing service providers, AI application enterprises, algorithm platforms, and internet platform enterprises will all be driven forward by the development of generative AI.
It is worth mentioning that in the era of AI, East-West computing holds greater significance. Guosen Securities believes that the unique resource endowments in the western region will become the most precious “cost-reducing” resource in the AI era. The East-West computing project makes it possible to reduce “data storage costs” to “reduce overall social operating costs”.
Pan Helin pointed out that generative AI requires intelligent computing centers, and intelligent computing centers may need to be closer to the demand side to reduce latency. Therefore, on the one hand, East-West computing provides computational support for generative AI, effectively promoting its development. On the other hand, as the characteristics of intelligent computing power change, the East-West computing layout will also adjust and distinguish between data centers and intelligent computing centers.
How is the IGC field regulated?
On March 31, according to CCTV, the Italian Personal Data Protection Bureau stated that it had launched an investigation into OpenAI’s chatbot ChatGPT for suspected violations of data collection rules. From now on, the use of ChatGPT is prohibited, and OpenAI’s processing of Italian user data is temporarily restricted.
In addition, the European law enforcement agency, Europol, has warned that ChatGPT may be misused for phishing, false information, and cybercrime; the non-profit organization, Center for AI and Digital Policy (CAIDP), has also lodged a complaint with the US Federal Trade Commission (FTC) stating that GPT-4 “is biased, deceptive, and poses risks to privacy and public safety.”
In China, AIGC (using AI technology to generate content) also has many risks, including privacy risks, misuse of AI-generated content risks, data missing or incorrect risks, intellectual property risks, and so on.
Recently, the China Payment and Clearing Association issued an initiative on “prudent use of ChatGPT and other tools by payment industry practitioners,” which has attracted industry attention. The association called on the industry to raise awareness, use AI technology in accordance with laws and regulations, not upload key sensitive information, and strengthen internal management and guidance of member units.
However, the existing regulatory system in China has already taken shape with regard to the risks posed by AIGC. At the level of laws and regulations, on the one hand, specialized comprehensive legislation regulates the obligations and responsibilities of network operators in the use of AI technology, and on the other hand, specific regulations are focused on algorithms, models, and technology in the field of AI.
On April 11, in order to promote the healthy development and standardized application of generative AI technology, in accordance with the “Cybersecurity Law of the People’s Republic of China” and other laws and regulations, the National Internet Information Office drafted the “Management Measures for Generative AI Services (Consultation Draft)” and solicited opinions from the public. The consultation draft regulates aspects such as personal information protection, legality of data sources, protection of intellectual property rights, and constraints on unfair competition behavior.
The consultation draft also proposes that generative AI providers should label generated images, videos and other content in accordance with the “Deep Synthesis Management Regulations for Internet Information Services”. Previously, the regulatory scope of the “Deep Synthesis Management Regulations for Internet Information Services” was only a reference for the standardization of AIGC technology and services.
Beijing lawyer Zhao Dong said in an interview with China Net Finance that in the future, the “three laws on the internet” (“Cybersecurity Law”, “Data Security Law”, and “Personal Information Protection Law”) and the related systems and regulations they provide will still be an important basis for the regulation of new technologies. Strict regulation of the emergence and application of new technologies in accordance with the “three laws on the internet” is not only conducive to maintaining network security, data security, and personal information security, but also to the healthy and orderly development of the entire industry. Pan Helin believes that the release of the consultation draft mainly clarifies the overall tone of first regulating and then developing generative AI, and is generally aimed at improving the application environment of generative AI and creating conditions for its domestic use. However, from another perspective, the content of the consultation draft is still relatively simplified, and some of the clauses are still relatively coarse. He recommends that the content of the consultation draft needs to be further improved in the future to prevent situations where “one regulation kills, one release disrupts” in the field of generative AI and to find a balance between the development and safety of generative AI. “It is necessary to regulate AIGC, and the regulatory goal should not be the personality of AI, but the humans using AI and understanding what they want to achieve,” Pan Helin said frankly. Zhao Dong believes that the current regulatory measures, the basic principles, systems and requirements stipulated by the “three laws on the internet” are the “minimum requirements” for maintaining network security, personal information security and data security, and the relevant system and rules do not restrict the development of new technologies, but rather hope that new technologies can promote data-based digital economic development and empower economic and social development comprehensively.