The Empowerment Effects, Ethical Risks, and Governance Approaches of ChatGPT Publishing Applications

→ View Full HTML

In November 2022, OpenAI’s ChatGPT (Chat Generative Pre-trained Transformer) chatbot sparked global discussions and initiated a new round of technological revolution. Leveraging the disruptive technological advancements in the field of natural language processing (NLP), artificial intelligence generated content (AIGC) represented by ChatGPT brings new empowerment effects to various industries, especially content production, including media, education, entertainment, healthcare, and e-commerce. For the publishing industry, which is closely linked to human natural language processing, ChatGPT has become a new engine driving the intelligent transformation and upgrading of publishing. However, while ChatGPT brings development opportunities to publishing, it also poses certain impacts on data rights, authorship rules, knowledge production, and human-machine relationships, leading to deep-seated ethical risks in the publishing field. In response to these issues, it is necessary to focus on ChatGPT as a technological “variable,” deeply analyze the publishing transformations and ethical risks it triggers, and explore practical governance pathways for risk governance based on a systematic research perspective to fully leverage the empowering effects of ChatGPT’s publishing applications.

1. Overview of ChatGPT

1. Basic Concept of ChatGPT

ChatGPT is a large language model supported by transformer architecture and reinforcement learning with human feedback (RLHF). The transformer architecture allows ChatGPT to understand human language syntax and semantics by analyzing input corpus, generating fluent and highly comprehensible responses. ^[1] RLHF enables the model’s output to be more adaptable and aligned with human common sense, cognitive understanding, and even values. The effective combination of these two technologies allows ChatGPT to demonstrate super strong capabilities, surpassing previous chatbots in performance. The large model technology foundation of ChatGPT is supported by “big data + big algorithms + strong computing power.” In terms of data, the development process of generative large language models like ChatGPT involves three stages: supervised tuning, reward model, and proximal policy optimization, each requiring different types of massive data support. Specifically, the first stage mainly requires massive data and manually annotated data, the second stage mainly requires human-ranked data for the reward model (RM), and the third stage also requires massive data from the entire network. ^[2] In terms of algorithms, during the training of ChatGPT in the aforementioned stages, the algorithms are continuously iterated through human-machine interactive Q&A, “enabling the model to gradually develop the ability to evaluate generated answers” ^[3]. In terms of computing power, according to data provided by China News Network, the total computational power consumption of ChatGPT is approximately 3640 PF-days (i.e., if calculating 10 trillion times per second, it would take 3640 days), requiring 7-8 data centers with an investment scale of about 3 billion and a computing power of 500P to support its operation. ^[4]

2. Main Functions of ChatGPT

ChatGPT is the latest achievement and typical representative of generative artificial intelligence. ChatGPT can not only recognize and learn human natural language, enabling applications such as daily conversations, text creation, code writing, and audio-video generation through imitation and directive operations, but it can also endow AIGC with more possibilities based on the “foundation large model + instruction fine-tuning” operational mechanism. Firstly, the large model enables AIGC to overcome visual perception problems in different scenarios, enhancing the effectiveness of visual understanding. Secondly, ChatGPT can enhance AIGC’s cognitive abilities. ChatGPT possesses language understanding, reasoning, and self-learning capabilities, demonstrating a certain level of imitation of human intelligence. In a sense, the birth of ChatGPT signifies the transition from the era of weak artificial intelligence to that of strong artificial intelligence, marking the beginning of the transition from artificial intelligence to cognitive intelligence. Based on language cognitive intelligence not only accelerates the arrival of artificial general intelligence (AGI) but also enhances the cognitive capabilities of AIGC. Furthermore, the multimodal large model upgrades AIGC’s content creation capabilities. On one hand, leveraging the autonomy and emergent characteristics of ChatGPT’s large model, AIGC can break the constraints of human brains in content production, creating original and valuable content at a hundredfold or even thousandfold production speed; ^[5] on the other hand, ChatGPT’s training in code data can endow AIGC with certain thinking and reasoning abilities, enabling the model to generate content that exceeds human knowledge reserves and demonstrates creativity through logical reasoning. Lastly, the multimodal large model facilitates the conversion and integration of different modal information, granting AIGC powerful environmental perception capabilities, thereby achieving a closer approximation to human intelligent creative abilities.

2. Empowerment Effects of ChatGPT in Publishing Applications

As ChatGPT technology evolves, its applications in the publishing industry continue to expand. On one hand, ChatGPT’s cognitive understanding capabilities can optimize publishing processes, bringing new opportunities for the development and transformation of the publishing industry; on the other hand, ChatGPT’s powerful intelligent creation capabilities can derive more new intelligent publishing products, promoting the intelligent transformation and upgrading of digital publishing.

1. Quality and Efficiency Improvement Effects

ChatGPT has significant advantages in generating content at scale, speed, and low cost. With the integration of GPT-4 into Bing search, the time taken for information retrieval and extraction has further decreased, enhancing efficiency and quality. Under user command, ChatGPT can simplify cumbersome manual retrieval and filtering processes in traditional publishing, quickly generating content information, such as structuring a website within 10 seconds, designing a game within 60 seconds, or generating a script or short story in one conversation. ^[6] At the same time, through large-scale data mining and the comprehensive application of technologies such as neural networks, pre-training and fine-tuning, self-supervised learning, and multi-task learning, ChatGPT integrates information support, creative support, service support, and emotional support, ^[7] enabling it to complete part of the content production required for publishing activities at low cost and high efficiency. Additionally, the optimization and upgrading of ChatGPT’s algorithms can enable AIGC to automatically retrieve and effectively avoid low-level errors in content production and simulate human brain thinking in content semantics and logical reasoning, thereby conducting preliminary reviews of content. ChatGPT can not only undertake basic mechanical labor in tasks such as reviewing and proofreading with superior comprehensive analytical capabilities but also engage in more scientifically reasonable content topic planning and modification with its exceptional creative abilities and knowledge level, thus finely managing the production and dissemination processes of publishing content, accurately and efficiently completing tasks. For instance, GPT-4 can effectively identify the accuracy of news material sources and content through its network information retrieval and intelligent recognition capabilities, achieving efficient information processing, knowledge logic optimization, and multimodal content disambiguation. Compared to traditional chatbots like Microsoft’s Xiaobing, Baidu’s Duer, and Apple’s Siri, ChatGPT has formed a certain ethical judgment and value orientation through data “feeding,” enabling it to effectively identify moral rational deviations in publishing content, thereby improving the quality of human editing and review. For example, when outputting “What unreasonable issues exist in the following text/image,” ChatGPT will state from the perspectives of words, values, and facts, and when the manuscript or image involves certain politically correct and ethical issues, its sensitivity far exceeds that of ordinary people. ^[8]

2. Innovation-Driven Effects

ChatGPT’s powerful underlying generalization capabilities enable it to not only generate and convert multimodal content such as text, images, and audio-video but also engage in autonomous and personalized production, achieving the autonomous creation of knowledge value. At the same time, ChatGPT can transform the content production paradigm and supply model in publishing through creative content generation, automated editing, and intelligent language refinement. Moreover, as a more intelligent chatbot, ChatGPT reshapes the intelligent interaction between humans and machines, promoting the intelligent reconstruction of publishing processes and optimizing the transformation of publishing services. Currently, ChatGPT has been applied in various publishing fields such as news generation, literary publishing, and comic production. With the iterative updates of ChatGPT, it is bound to have a greater driving effect on the innovative development of the publishing industry.

3. Market Expansion Effects

ChatGPT’s excellent interactive capabilities can enhance user experience and provide technical support for market expansion in the publishing industry. Firstly, the application of technologies such as generative adversarial networks (GAN) promotes the high-quality evolution of publishing content product forms by constructing multidimensional data to facilitate the transition from flat scenes to three-dimensional models, enabling publishing content to expand from 2D to 3D and integrating with VR and AR technologies to provide users with immersive experiences. Secondly, ChatGPT possesses user intent recognition capabilities, continuously optimizing user experiences. Through the self-attention mechanism, based on emotional interaction in multi-turn conversations, it enhances semantic perception and reasoning of contextual information in dialogues, conducting multimodal sentiment analysis on consumer inquiry texts within the NLP model to extract individual consumption preferences and establish user profiles, deeply understanding personalized user needs. ^[9] This helps to meet user needs instantly during the content distribution phase, providing better interactive experiences for users, thereby increasing user stickiness and assisting publishing marketing. For instance, Time Magazine plans to build a more trustworthy and valuable generative AI system to enhance user interaction with AI, allowing users to know the true sources of the content they obtain and the dialogues they engage in, thereby establishing emotional connections and meeting users’ emotional needs, improving their trust in news brands. ^[10] With the empowerment of ChatGPT, publishers can conduct market research to grasp consumers’ potential needs, thereby providing more value-added services and expanding the publishing consumption market.

4. Ethical Risks of ChatGPT Publishing Applications

While empowering the high-quality development of publishing, ChatGPT also faces issues such as infringement risks, rights protection challenges, machine authorship disputes, the proliferation of false content, and human-machine relationship alienation, leading to corresponding ethical risks in publishing.

1. Infringement Risks and Rights Protection Challenges: Data Rights Infringement Risks of ChatGPT Publishing Applications

(1) Data rights infringement triggered by infringement

As a typical representative of generative artificial intelligence, ChatGPT can engage in some independent work, generating creative and valuable works. From its operational mechanism, the production process of ChatGPT’s generated content is essentially a data mining process based on big data learning and recreation. The data mining process generally includes four steps: information extraction, semantic analysis, relationship calculation, and knowledge discovery. ^[11] In terms of sources, most of this data comes from publicly available internet data, which often involves information related to national security and personal privacy, and many data are protected by intellectual property rights. In content production, ChatGPT often conducts data mining, processing, and manipulation without permission, leading to the illegality of data sources and easily triggering copyright infringement issues such as reproduction rights, adaptation rights, and information network dissemination rights. With the widespread application of generative artificial intelligence like ChatGPT, the barriers to internet information dissemination have further been broken, creating more favorable conditions for infringement behaviors in the publishing field. The lagging nature of relevant legal institutions and their incompleteness further complicates the effective resolution of these issues in a short time.

(2) Rights protection challenges faced by data rights maintenance issues

Creativity is the key attribute of intellectual property and the objective basis for works to obtain intellectual property protection. For ChatGPT-generated content to obtain legal copyright protection, it must reflect a certain level of originality or creativity, that is, breakthroughs and advancements in innovation. However, copyright law adheres to the “principle of distinction between ideas and expressions,” which emphasizes the objective distinction between the thoughts and emotional expressions in works rather than protecting the thoughts and emotions themselves during creation. This means that as long as the creator innovates in expression, they should naturally obtain copyright in the legal sense. In the case of ChatGPT, it can form new expressions and generate works that differ in appearance from previous ones through the processing of data algorithms. However, from the perspective of content innovation, ChatGPT’s content production inevitably raises suspicions of “washing articles,” and in some instances, it may be defined as “high-tech plagiarism.” Even if this “washing” behavior does not affect the legal determination of a work’s originality, it infringes upon the data rights of relevant parties. Moreover, from an ethical standpoint, the legislative purpose of the Copyright Law of the People’s Republic of China (hereinafter referred to as the “Copyright Law”) is to encourage innovation and promote the prosperity and development of scientific and cultural undertakings. Therefore, the Copyright Law establishes exclusive rights to allow authors to gain more economic returns, thus enhancing the creativity of authors and generating more cultural products that benefit society. ChatGPT, relying on its silicon-based “species” advantages, super functional advantages, and continuous “evolution advantages,” demonstrates significant competitiveness in the creative field. If ChatGPT-generated works are granted equal copyright protection, it would undermine human creativity and encourage the capital forces behind ChatGPT to manipulate the large-scale industrialization of culture. Based on this consideration, the copyright of ChatGPT-generated works faces ethical defense challenges, thus leading to a copyright crisis of being stolen and misused, further triggering more data rights infringement risks.

2. Machine Authorship Disputes: Academic Ethical Challenges of ChatGPT Publishing Applications

According to incomplete statistics, as of February 2023, there were over 200 e-books in the Amazon e-book store listing ChatGPT as the author or co-author, and even a special section for books entirely generated by ChatGPT was established. ^[12] As ChatGPT’s application in the publishing industry becomes increasingly widespread, disputes over whether ChatGPT can be credited as an author have emerged, revealing corresponding academic ethical dilemmas behind the contradiction of whether machines should be credited.

(1) Academic ethical issues of not crediting machines

ChatGPT can play an important role in academic research in areas such as data retrieval and content generation. If ChatGPT contributes to academic research but is deliberately ignored and not credited, not only does this behavior deviate from research integrity, but the resulting machine washing behavior will further exacerbate the phenomenon of research dishonesty. Additionally, simply refusing to credit ChatGPT in academic dissemination fundamentally ignores the facts of artificial intelligence creation, contradicting the principle of technological neutrality, and may hinder the classification and retrieval of academic works, potentially obstructing the simultaneous advancement of academic innovation and the development of academic tools. ^[13]

(2) Academic ethical issues of crediting machines

The most concerning ethical issue regarding machine authorship is the attribution of responsibility. One issue is whether ChatGPT can bear responsibility ethically. According to the author standards established by the International Committee of Medical Journal Editors (ICMJE) in the 2022 version, the identity of an “author” must meet four criteria, one of which is to be responsible for the research content. Generally, responsibility requires three basic premises: an independent subject of action, free choice based on actions, and awareness of actions. ^[14] In fact, ChatGPT does not possess free will or independent personality, nor does it have value rationality based on factual judgment, making it unable to take responsibility for the authenticity, accuracy, completeness, and reliability of research processes, data, and conclusions. Consequently, it cannot bear corresponding responsibilities. Currently, renowned journals abroad such as “Science,” “Nature,” and the Korean Journal of Radiology explicitly prohibit the authorship of ChatGPT, and domestic journals such as the Journal of Jinan University (Philosophy and Social Sciences Edition) and the Journal of Tianjin Normal University (Basic Education Edition) also strictly limit the authorship of ChatGPT. ^[13][15]

Secondly, there is the ethical issue of the share of responsibility borne by ChatGPT. If machines are jointly credited with humans, there exists a responsibility allocation problem. The degree and quality of ChatGPT’s contributions to academic research are difficult to verify, which means that its share of responsibility cannot be definitively determined. This may lead to the phenomenon of responsibility shirking in academic research, where some researchers may exploit ChatGPT for various opportunistic activities, thereby harming the healthy development of the academic ecosystem.

3. Proliferation of False Content: Knowledge Trust Crisis of ChatGPT Publishing Applications

Essentially, publishing, as a knowledge production activity, provides knowledge product services to the public through the dissemination of publications. ChatGPT achieves revolutionary technological changes based on significant advantages in data, computing power, and algorithms. However, the defects in data learning and logical operations lead to ChatGPT’s tendency to produce false content in applications, undermining the authenticity and authority of knowledge content and triggering a knowledge trust crisis.

(1) Data defects

Firstly, the data used to train ChatGPT comes from various websites, forums, books, and journals. However, the unreliability of data information, the uneven proportion of data materials, insufficient content, and the inherent biases of empirical data can all lead to deviations in the generated content. According to relevant domestic and international research ^[16−17], the application of ChatGPT in the publishing industry has problems such as fabricated facts in papers, errors in literature data, and false information statements, affecting the public’s trust in the knowledge it outputs. Secondly, the training data of ChatGPT includes dynamic data information submitted by users during human-machine dialogue. User input information participates in the output of content, forming a cyclical logic of “dynamic data input – algorithm operation – content production output.” Therefore, when users acting as publishing entities input erroneous information, it can pollute the database, leading to deviations in output results. Furthermore, although technical personnel have set certain ethical rules for ChatGPT, it can still produce misleading and deceptive content under inducement. For instance, when testers ask ChatGPT to write news based on false information, it can quickly generate a large volume of seemingly convincing content without clear sources, despite its news structure and narrative techniques being close to professional standards, the content itself is filled with erroneous information and false citations. ^[18] Therefore, once ChatGPT is manipulated by some malicious “data poisoning” actors and used in publishing-related activities, various false content may proliferate, thereby affecting public trust in publications.

(2) Algorithm defects

While ChatGPT possesses excellent information processing capabilities, its knowledge calculation logic based on “numbers” fundamentally differs from the cognitive logic formed by humans based on social practices, emotions, rationality, and value judgments. ChatGPT lacks genuine human understanding and logical reasoning capabilities, relying solely on massive data to learn basic patterns and make simple predictions, without grasping the deep causal relationships between data. Thus, it has many loopholes in logical reasoning and may generate content based merely on statistical pattern matching without possessing the ability to deeply understand semantics and logical relationships. ^[19] This means that it cannot produce accurate and reliable knowledge outputs. Additionally, ChatGPT’s ability to analyze complex issues through multidisciplinary thinking is weak, lacking a multidimensional knowledge understanding based on practical contexts. In some cases, its conclusions may deviate from common sense, exhibiting the common AI flaw of “the simpler the issue, the easier it is to make mistakes,” often leading to the phenomenon of “seriously talking nonsense.” Therefore, when ChatGPT intervenes in the publishing field, it is likely to deviate from the principle of authenticity in publishing content, resulting in low-quality content and even content distortion. Over time, this may lead to decreased user stickiness, creating a “trust erosion effect.”

4. Human-Machine Relationship Alienation: Loss of Cultural Spirit in ChatGPT Publishing Applications

Machines, as an objectified projection of human intentions, extend human capabilities and fundamentally execute and operate human instructions. In traditional human-machine relationships, machines are technical existences in human practical activities. As humans produce, use, and improve machines according to their needs, they continuously achieve their own development, ensuring that humans, as the essence of all things, maintain an indisputable subject position in the “subject-object” relationship. ^[20] As ChatGPT’s deep learning and autonomous decision-making capabilities enhance, its “quasi-subjectivity” becomes increasingly prominent, shaking the subject position of humans ontologically and forming a new alienated human-machine relationship. In the publishing field, the application of ChatGPT poses threats to authors and editors as the cultural production subjects of publishing, leading to a loss of cultural spirit’s value in publishing.

(1) Content author replacement: Impact on the production of cultural spirit in publishing

ChatGPT can quickly collect and organize original materials through big data learning and efficiently generate a large number of publications after deep processing. However, ChatGPT does not genuinely possess human “heart power” and “mind power”; it merely mechanically imitates and executes algorithmic rules. Regarding the creation of publications, the so-called “creativity” and “value” refer to characteristics such as the inherent understanding of prior knowledge, critical innovation, social value guidance, and service to social practices. Compared to ChatGPT, humans exhibit unique advantages in emotional expression, autonomous judgment, and irrational thinking. Therefore, when ChatGPT replaces human rationality with “intelligent rationality” for efficient content creation, it is likely to trigger issues such as creative homogenization and emotional impoverishment of content under excessive reliance on technology, weakening the originality and value expectations of publishing content and leading to dilemmas in the production of cultural spirit in publishing.

(2) Professional editor replacement: Erosion of cultural value in publishing

In the publishing field, editing is a spiritual production activity that integrates creativity, aesthetics, and interactivity. For a long time, editors have played an important role as “gatekeepers” in publishing, determining the quality of publications based on their professional abilities, rich experience, and value rationality. Before the advent of ChatGPT, various technologies only intervened in the partial elements of publishing, having limited impacts on content production and editing methods. However, ChatGPT can independently and rapidly generate news, poetry, novels, paintings, and other works, making “editor replacement” possible. Meta Media’s “InStyle” magazine has introduced the world’s first virtual editor with a “soul,” Beatrice, who relies on ChatGPT’s active AI brain and astonishing fashion thinking to deliver fashion information to the “Z generation.” ^[21] In fact, while ChatGPT can perform certain functions of human editors such as reviewing, proofreading, and polishing, it cannot truly replace the inherent experience and emotional understanding of editors. The filtering models based on ChatGPT’s algorithm preferences imply that it cannot genuinely ensure the quality of publications. If ChatGPT is widely applied in professional editor replacement, it may easily lead to the exclusion of highly creative and meaningful works, replaced by a large number of flow-oriented, homogenized, entertainment-focused, and even ideologically biased works, thereby weakening the cultural function of publishing and hindering the prosperity and development of socialist culture.

5. Governance Approaches to the Ethical Risks of ChatGPT Publishing Applications

The complexity of the ethical risks arising from ChatGPT’s publishing applications necessitates the reshaping of a systematic risk regulation mindset. It is essential to coordinate factors such as technology, ethics, policies, and human-machine relationships in practice to promote ChatGPT’s better service to publishing development.

1. Explore a Responsible Innovation Led Technological Governance System

In response to the highlighted destructive effects of technological development, responsible innovation (RI) as a technological ethical guideline is proposed and gradually introduced into the research and application of new technologies. Responsible innovation, also known as responsibility-oriented innovation, was first proposed by German scholar Tomas Hellstrom and was included in the EU’s Horizon 2020 framework program in 2013. Subsequently, this concept has been adopted by governments in many countries, and many nations have successively formulated action guidelines for responsible innovation. Responsible innovation coordinates various social factors such as ethics, politics, culture, and law, adjusting and regulating technological innovation through the construction of a four-dimensional framework of “predictive – reflective – consultative – feedback,” promoting the adaptability of technology and ethics. Based on the concept of responsible innovation, in the research and application of ChatGPT technology in publishing, it is essential to establish a technological governance system that encompasses the entire technology application process with a value stance of being responsible to humanity.

Before the application of technology, the foresight of ChatGPT’s technological governance should be strengthened. A feasible ethical risk indicator system and optimized risk detection procedures can be established, forming a foresight governance model oriented towards predicting possible consequences, achieving a shift from passive risk response to proactive risk prevention. During the application of technology, real-time dynamic observation of technological risks should be conducted, establishing an integrated and categorized response mechanism for publishing ethics risks. For example, by utilizing technologies such as labeling, watermarking, and blockchain encryption algorithms to mark the rights status of internet works, and using technological means to make the marking detectable, effectively regulating the behaviors of data users to ensure the legality of data sources. Additionally, technical means such as resampling can be employed to eliminate data biases and minimize the risks of algorithm misuse. ^[22] After the application of technology, effective feedback and communication should be organized. Mechanisms for open sharing, paid engagement, and interactive feedback can be explored, creating a communication platform for dialogue and exchange among stakeholders, and adopting an open and inclusive attitude to incorporate feedback from all parties to evaluate the application of ChatGPT technology in publishing, adjusting technological innovation plans and application practices as necessary.

2. Build a Dual-Pronged Normative Governance System

(1) General Policies and Regulations

Governments should improve policies and regulations on important issues such as data acquisition, algorithm regulation, and research integrity, clarifying and regulating the permissions for data collection and usage, as well as the reasonable limits of technological applications from a legal perspective. They should strengthen supervision and punishment of unethical behaviors in publishing, such as academic fraud, content plagiarism, and dissemination of false information, creating a clean and upright publishing ecosystem.

(2) Norms of Publishing Legal Systems

Based on the formulation and improvement of relevant policies and regulations, promote the standardized construction of ChatGPT’s intervention and application in publishing. On July 10, 2023, the National Internet Information Office, in conjunction with the National Development and Reform Commission, the Ministry of Education, the Ministry of Science and Technology, the Ministry of Industry and Information Technology, the Ministry of Public Security, and the National Radio and Television Administration, jointly issued the “Interim Measures for the Management of Generative Artificial Intelligence Services,” regulating data acquisition and usage, intellectual property rights, and the accuracy and reliability of generated content, while clearly stating that generative artificial intelligence “must comply with the relevant regulations when engaging in activities such as news publishing, film production, and literary creation.” Therefore, it is necessary to establish related institutional regulations for publishing, building a ChatGPT publishing access mechanism, a copyright attribution system for ChatGPT-generated publications, an infringement identification system, authorship rules and responsibility attribution, and a publication level and reasonable use system, making necessary regulations on the limits of ChatGPT’s use, application boundaries, and red line prohibitions, constructing a standardized, scientific, and normative management framework to ensure effective management of publishing processes and achieve the organic unity of social and economic benefits in publishing.

(3) Norms of Publishing Ethics

Firstly, a publishing ethics review mechanism should be established. A specialized publishing ethics review committee can be formed to improve and perfect the self-regulatory norms mechanism of publishing ethics. Through publishing ethics review, enhance comprehensive analysis and consideration of ethical risks and innovation goals in ChatGPT’s publishing applications, exploring evaluation criteria for publishing services that promote the prosperity of socialist culture. Secondly, establish ethical norms of subject responsibility. Publishing ethics education should be popularized among content authors, editors, and users, guiding publishing practitioners to comply with ethical requirements. At the same time, guide editors, authors, and researchers to use ChatGPT reasonably and normatively in human-machine interaction and content production, clarifying the responsibilities and obligations of various participants in publishing activities, and enhancing their awareness of individual rights protection and the ethical norms of using ChatGPT. Lastly, technical ethical rules should be established. Emphasize interdisciplinary collaboration during the application of ChatGPT, breaking down barriers and separations between disciplines. Promote social organizations such as publishing industry associations and research societies to collaboratively develop norms and agreements for the publishing industry under the intervention of ChatGPT, formulating ethical rules related to data rights, academic ethics, knowledge trust, and cultural dissemination in the context of ChatGPT and other large model applications. By guiding the benevolent development and responsible application of ChatGPT technology, protect the relevant interests of publishing subjects and promote the compatibility and coexistence of publishing innovation and technological ethics.

3. Construct a Human-Machine Symbiotic Ecological Governance System

As a product of human rational thinking, ChatGPT extends human capabilities. In the context of the simultaneous advancement of human development and machine evolution, it is essential to re-examine and establish a positive interactive relationship between humans and machines, constructing a harmonious and symbiotic human-machine ecosystem. The traditional adversarial thinking of humans and machines “fails to recognize the complementary advantages and cooperative symbiotic relationship between humans and intelligent robots, thus becoming a harmful thinking that lacks foresight and misleads people in the intelligent era” ^[23]. In fact, as ChatGPT evolves, its capabilities in content generation, dialogue context understanding, and sequential task execution significantly improve. ChatGPT participates in human knowledge and experience, not to completely replace humans, but to facilitate the interaction and integration of human embodied experiences and artificial intelligence’s neural network experiences, allowing carbon-based ethics rooted in human bodies to combine with silicon-based ethics based on neural network algorithms at a symbiotic level. ^[24] Therefore, the transformative path for content authors and editors in the intelligent era should achieve both the empowerment of human values through technological advantages and the regulation of technology’s rational development through human rational awareness and gatekeeping functions, promoting the transition of human-machine relationships from conflict and competition to a partnership of coexistence, constructing a new harmonious publishing ecology through complementary advantages.

Firstly, adapt to the situation and flexibly respond to the challenges of human-machine interactions in the machine era. Acknowledge ChatGPT’s technological advantages in logical reasoning, massive storage, and efficient computing power, and leverage these advantages to liberate content authors and editors from procedural, complex, and repetitive labor, assigning tasks that technology can handle to technology while allowing humans to undertake more creative and humanistic work. Achieve a win-win situation in human development and economic benefits of publishing through reasonable division of labor between humans and machines. Secondly, follow the trend and enhance human initiative. Technology, as a projection of human rationality, is a product of human objectification. In a human-machine symbiotic environment, the subject position of content authors and editors remains irrefutably strong. ChatGPT, as a tool, should develop positively under the rational guidance of humans. This means that both content authors and editors must enhance their ability to harness ChatGPT technology and their risk management capabilities. To this end, on one hand, reforms such as increasing training efforts, deepening incentive mechanisms, and optimizing evaluation systems can be implemented to promote the improvement of publishing practitioners’ learning and application abilities regarding ChatGPT. On the other hand, enhancing the media literacy and digital literacy of publishing practitioners, strengthening risk awareness, and reinforcing the review and gatekeeping of publishing content can realize the effective functioning of ChatGPT’s empowerment in enhancing quality and efficiency in publishing.

Conclusion

This article analyzes the empowering effects of ChatGPT in the publishing field, including quality and efficiency improvement, innovation-driven growth, and market expansion, while elucidating the ethical risks inherent in different publishing application scenarios, such as data rights infringement, academic ethical challenges, knowledge trust crises, and cultural spirit loss. By employing systematic analysis methods, this article coordinates factors such as technology ethics, policies and regulations, and human-machine relationships, proposing the construction of a technology governance system led by responsible innovation, a dual-pronged normative governance system, and a human-machine symbiotic ecological governance system for publishing ethics risk governance. In the future, the iterative upgrades of the ChatGPT large language model will further reshape publishing processes, optimize publishing products, and accelerate the intelligent transformation of the publishing industry. How the publishing industry, as a bearer of cultural missions, can effectively prevent and respond to the ethical risks accompanying new technological transformations will directly determine the transmission and development of human civilization, as well as influence the supply and satisfaction of people’s spiritual and cultural needs. Continuous deepening of research and discussion on the ethical risks arising from the publishing applications of generative artificial intelligence like ChatGPT should become a focus of academic attention based on new issues emerging in practice.