Detailed replies from ChatGPT on specific questions in the Chinese language Photo: CFP 

Beijing Normal University recently hosted an online salon on “Viewing ChatGPT from Journalism and Communication.” What breakthroughs will ChatGPT bring from the perspective of natural language models? Will it bring about a revolution in journalism and communication? What opportunities and challenges will it bring? Participating scholars from natural linguistics, journalism and communication, and computational communication exchanged views on these questions. 

Changing ecologies 

ChatGPT is the latest chatbot model developed by the American AI research company OpenAI. Impressively, only two months after its launch, the number of active users exceeded 100 million. Along with a strong learning capacity, it also possesses a high degree of “intelligence.” It can not only chat with people, but also answer various complex professional questions. As an AI-powered natural language processing tool, ChatGPT conducts conversations by learning and understanding human speech, while interacting according to specific chat contexts. 

Why has ChatGPT gone viral so quickly? Huang Minlie, an associate professor from the Department of Computer Science and Technology at Tsinghua University, said that AI has entered the era of humanoid interaction, generating dialogue models close to human levels. As a general task assistant, the ChatGPT model includes some non-traditional open-ended tasks. High-quality dialogue gives the false impression that the AI has consciousness and an awakened personality. A harmonious “AI-human” existence is an inevitable social trend. 

According to Wu Ye, director of the Center for Computational Communication Research at Beijing Normal University, ChatGPT is a big step forward in improving AI technology’s ability to process human language, but it’s still too early to consider it an AI revolution. It merely extracts information from a massive knowledge base without rational thinking. Professional topics are often answered with loopholes, thus requiring multiple rounds of professional close interaction between human and machine for continuous guidance and improvement. 

“The advances of AI have mainly led the natural language processing technology,” said Xu Xiaoke, a professor from the School of Information and Communication Engineering at Dalian Minzu University. Although ChatGPT cannot be counted as a breakthrough technology, it still represents a landmark AI application. For instance, in the field of journalism, it can be applied to the coverage of daily sporting events or financial and stock market reports to effectively reduce costs. It uses AI to generate content, altering the relationship between human and social bots and the ecology of coexistence of human and AI. 

New research topics 

ChatGPT merges with the identity of the communication subject, demonstrating the importance of AI content generation and dissemination. It is being embedded in people’s daily lives due to its high degree of socialization. Zhou Baohua, a professor from the School of Journalism at Fudan University, said that for empirical research on journalism and communication, the use of AI technology for content generation and transmission is still in its infancy. Compared with previous mainstream applications, ChatGPT has moved from interpersonal communication to public communication, and has increased in both scale and public influence. 

Regarding ChatGPT’s impact on journalism and communication, Zhou noted that the development of human knowledge is tightly linked to media. In the face of a new knowledge medium, we should first consider how new knowledge is generated in the manner of AI-generated content (AIGC), while examining the relationship between certainty and uncertainty of machine generated content. Efforts are also needed to reveal the process of power operation and social construction behind ChatGPT. As knowledge production goes through networked public access and circulation, it calls for special attention from journalism and communication scholars. AI, represented by ChatGPT, should be treated as a real issue in communication studies. 

According to Zou Jun, a professor from the School of Journalism and Communication at Guangzhou University, the biggest difference between ChatGPT and other search engines is that it can provide inspiration and creative generation, implying changes in the holistic information collection methods of communication studies. In the future, the application of AI in the journalism and communication industries may produce “Q&A style” news, that is, under certain circumstances integrate and generate information in light of the content that people want to know. With a high level of intelligence, this type of generation enables a personalized news service and reforms the news industry. In this process, more attention should be paid to the ethical issues of communication, striving to transcend technical logic and return to humanity. 

Risks and challenges 

Though the emergence of ChatGPT is hardly surprising amidst the rapid AI development, we still seem unprepared. Speaking of the risks and challenges of ChatGPT, Zhang Hongzhong, director of the Research Center for New Media Communication at Beijing Normal University, compared it to the days of horse-drawn carriages, when we could not predict the potential dangers of automobiles or make rules for them. The emerging ChatGPT should be judged by the logic of technological development. ChatGPT is a breakthrough result of improved computing power. Meanwhile, AI including ChatGPT, is only a technical means, representing a more efficient and convenient technology and a technological iteration. 

“ChatGPT still has many limitations at the current stage, including a lack of factuality, generative illusion, an inclination to generate lengthy replies, and relative conservativeness in active interaction,” said Zhang Weinan, a professor from the Faculty of Computing at Harbin Institute of Technology. 

Min Yong, an associate professor from the School of Journalism and Communication at Beijing Normal University, argues that the disruptive impact of recommendation algorithms will be more profound than that of ChatGPT, which still provides the Q&A mode of information distribution, while the former has already leveraged the AI technology to infer people’s thoughts and motives to a degree. 

“As AI makes continuous progress, the field of journalism and communication will likely incorporate the investigation of the effect of machine generation and dissemination,” said Shen Fei, an associate professor from the Department of Media and Communication at the City University of Hong Kong. Shen suggested being vigilant against ChatGPT’s “seemingly serious nonsense” on some occasions, which might result in factual errors, knowledge blind spots, and deviation from common sense. 

Edited by YANG LANLAN