Chen Bing: The development of ChatGPT has entered the normal, and standardized innovation is the key

China-Singapore Jingwei, September 12 Topic: The development of ChatGPT has entered the normalization, and standardized innovation is the key

The author: Chen Bing, Deputy Dean of the Law School of Nankai University, Special Researcher of the China New Generation Artificial Intelligence Strategic Development Institute

According to the latest data released by website traffic monitoring service Similarweb, the number of website visits to the artificial intelligence chatbot ChatGPT continued to decline in August 2023, which is also the third consecutive month of decline in website visits since OpenAI launched ChatGPT in November 2022. Worldwide desktop and mobile visits to ChatGPT sites fell 3.2% to 1.43 billion in August, with both June and July falling by about 10%, the data showed.

How to view the phenomenon of ChatGPT's popularity fading?

ChatGPT is a sign that the development of artificial intelligence has entered a new climax, once launched, it has received great attention, and now in the face of the phenomenon of ChatGPT popularity gradually decreasing, on the one hand, ChatGPT popularity is a normal market phenomenon, in line with the general law of economic development. In the wave of the Internet, every hot spot and every wave of traffic is cyclical, and the so-called ChatGPT from "limelight" to "heat gradually receding", in fact, more from the peak state to normal. The reason why the initial ChatGPT increased the number of visits and increased the number of visitors was because most users visited the product out of curiosity, and the media propaganda and propaganda, it was logically unsustainable.

At the same time, because ChatGPT technology is not mature enough, the context recognition ability is not ideal, and the academic identification ability is not high, some people in China even use people's social control psychology to harvest the "IQ tax" to earn quick money in an endless stream, including paying for the purchase and sale of "usage times", spending money to buy and sell various "use instructions", high-priced trading, etc. The emergence of ChatGPT has brought many new problems and new phenomena, but it is often not technical. To a certain extent, these problems are due to some social phenomena spilled over after the explosion of ChatGPT, but under the gimmick of a new business model, they pretend to play a new round of electric fraud, and use consumption hotspots to make profits. After rapid growth, cooling is an inevitable phenomenon.

On the other hand, the phenomenon of the fading popularity of ChatGPT should be examined from a comprehensive perspective, and the advanced and progressive nature of ChatGPT should not be denied due to the hot and cold changes of its technology, and the innovative development potential of artificial intelligence should not be ignored because of short-term traffic. The reason why the AI wave is so surging is that the technical driving force behind it is the development of many parallel fields such as big data, computing power, and mobile Internet, and the disruptive breakthroughs in these parallel fields can play a crucial supporting role in supporting and promoting AI technology. The gradual decline of ChatGPT popularity is temporary, but the innovation and development of artificial intelligence is a long-term process, and all sectors of society still need to maintain a high degree of attention to this.

In addition, the accelerated introduction of relevant artificial intelligence legislation in relevant countries and regions has also set reasonable and effective rules and constraints on the development and application of generative artificial intelligence technologies and products represented by ChatGPT, especially requiring the providers of generative artificial intelligence services to bear corresponding legal obligations on the source of the training data they use to avoid the risk of infringement, which objectively affects the development and use of ChatGPT products.

The rise of domestic large models is possible?

Similarweb said in the report that ChatGPT was a near monopoly in the field when it was released, but by now, there are 9 chatbot applications comparable in size to ChatGPT. There is a view that the cooling of ChatGPT is also related to the decline in its capacity. ChatGPT visits have declined, and the rise of domestic large models may usher in the possibility.

At present, the era of AI big models is still in the early stages, ChatGPT is only a type of general artificial intelligence development, yet to be further developed, its potential technology monopoly, capital expansion, data abuse, algorithm discrimination, privacy infringement and other legal risks, may have a negative impact on market competition, enterprise innovation, consumer welfare or even national security, its actual performance is still far from meeting people's expectations, which also provides development space for domestic large models, pointing out the development direction.

According to statistics, 11 large domestic models have successively passed the "Interim Measures for the Management of Generative Artificial Intelligence Services", of which 5 in Beijing, 3 in Shanghai are the first to go online, 2 in Guangdong and 1 in Anhui are also opening one after another. The "quantitative change" in the scale of model parameters is driving the "qualitative change" in the function and performance of artificial intelligence. The newly launched general large model in China is more like an infrastructure, on which more vertical models in professional fields can be incubated to serve the digital transformation of all walks of life.

For example, Baidu's "Wenxin Yiyan" has broad development prospects, and its full opening is precisely from the use side to help the R&D side accelerate iterative optimization. Baidu and other Chinese AI companies will further develop generative AI in the direction of production tools, becoming a shock force for the deep integration of artificial intelligence and the real economy, which will help make advanced productivity such as AI, big data, and cloud services better serve China's modernization in the future. It should be noted that the current related industries are still in the early stage of development, and the input cost of their general large models is very high, and they also need to be effectively combined with specific application scenarios to continuously innovate and progress.

At the same time, we should also be soberly aware that for the development, training and launch of large models, facing the generation, expansion and use of the basic corpus, we must do a good job in security precautions, especially the selection and training of language data related to ideology, objectively speaking, 90%-92% of the global large model corpus is English as a data carrier, and the proportion of Chinese in the corpus is very low, which is a great challenge for the development and use of large models in the mainland, and the next step of cross-border data connectivity is imperative. Then, how the mainland responds to data security and the overall national security caused by the cross-border transmission of data is particularly worthy of attention in the development of domestic large models. In addition, it is also necessary to avoid rushing and causing concept hype.

When will the era of AGI (General Artificial Intelligence) arrive?

In the past nearly 60 years, the development of artificial intelligence technology and application itself has experienced three ups and downs, and the fundamental reason is that it has not produced AI products that can truly change the world. What we need is an intelligence that can think, communicate, and have emotions like humans, but far surpasses human intelligence in memory, calculation, reasoning, etc. At the same time, it is necessary to remain loyal and be able to operate in an orderly manner within the framework of legal rules and ethics. However, after so many years of development, they have been hovering in the stage of weak AI. Therefore, the technological progress of ChatGPT has brought great surprises to people, but there are also many limitations, and there are still many challenges in the era of AGI, especially the control and supply of algorithms and computing power, and the integration with the current law and ethics of human society.

Chatbots are just the starting point of AI, intelligence is much more than verbal ability, the basic element of biological intelligence lies in the ability of animals to interact sensorimotor with the world. In the future, AI should conduct multimodal interaction with the environment through multi-modal interaction with the environment through embodiment, such as listening, speaking, reading, writing, thinking, manipulating objects, and acting, to help machine intelligence transcend the limitations of text single modality and better help humans.

ChatGPT technology can help companies innovate business models, and with the support of ChatGPT, companies can provide tailored products and services to each customer through in-depth analysis of customer behavior and feedback. Enterprises can also gradually learn user needs based on user feedback and provide more personalized services, thereby improving user satisfaction. It is foreseeable that with the support of ChatGPT technology and products, more digital intelligent artificial intelligence products with user-centered deep interaction will continue to emerge, and the distinction between the B side and the C side will no longer be so obvious.

At the same time, ChatGPT's intelligent analysis capabilities can provide decision-making support for enterprises. In short, with the continuous advancement of algorithm technology and computing power technology, ChatGPT will further move towards a more functional version, which will be applied in more and more commercial fields to better serve human production and life. (Zhongxin Jingwei APP)

This article is selected and edited by the Zhongxin Jingwei Research Institute, and the copyright of the works produced by the selection is all reserved, and no unit or individual may reprint, excerpt or use it in other ways without written authorization. The views expressed in the selected content are only those of the original author, and do not represent the views of CSJW.

Responsible editor: Sun Qingyang