|
|
|
| The Technological Characteristics, Evolution Process, and Governance Strategy of ChatGPT Applications from the Perspective of Resilience |
| Chen Zhi |
| (School of Public Administration, Central South University, Changsha 410075, China) |
|
|
|
|
Abstract The new-generation artificial intelligence product ChatGPT from OpenAI has sparked heated discussions around the world. ChatGPT is considered to have opened up a new era of artificial general intelligence (AGI), and it is a crucial step towards moving from "weak AI" to "strong AI". Although there is great room for imagination about the future of ChatGPT, and some experts even believe that it will completely change the human world, any leapfrog progress is used to come with considerable risks. Therefore, to achieve a balance between technological innovation and public value and between promoting development and ensuring security in the application of ChatGPT, it is urgent to integrate theoretical perspectives and policy contexts for further exploration. It can be confirmed that the social risks of artificial intelligence are not only caused by technology itself but also exposed after being integrated and resonated with the social structure. Therefore, the research on ChatGPT should return to the practical applications, and systematically explore the theoretical logic and development strategies behind it. This study aims to answer the following questions: What are the academic characteristics of the ChatGPT's applications? What kind of evolutionary development processes in ChatGPT applications are presented? What should be done to systematically regulate and govern ChatGPT applications, and can a comprehensive research framework be thus constructed?#br#According to the general research framework of system theory, this study attempts to extract three key topics of ChatGPT applications, namely "technological characteristics", "evolutionary process", and "governance strategy", from the "physical layer", "operational layer", and "strategic layer", and then systematically discuss ChatGPT with resilience and its extended research paradigms of "evolutionary resilience" and "resilience governance". Firstly, it is clarified that "resilience" is not only a fundamental symbol of static capability but also a powerful manifestation of dynamic processes, which can be embedded as an effective interpretive tool in ChatGPT applications. Furthermore, by combining authoritative data and the latest cases, the study summarizes the characteristics of resilience in ChatGPT applications in the "To B" level of development compatibility and openness, as well as in the "To C" level of learning adaptation and humanoid adjustment. Secondly, it introduces the "exploitation r-conservation K-release Ω-reorganization α" of evolutionary resilience system, and elaborates on the different stages of evolution in ChatGPT application expansion, including rapid extension, stable adjustment, risk response, and comprehensive improvement. It is also emphasized that the "four stages" may also be staggered and overlapping, and may not necessarily evolve and develop in a linear order. Therefore, it is necessary to analyze specific situations and provide effective guidance for the systematic construction of resilience governance. Next, in the interpretation of the latest Interim Measures for the Management of Generative Artificial Intelligence Services, the principles of "inclusiveness and prudence" and the purpose of "responsible technological innovation" are confirmed to be the core principles for the sustainable applications of ChatGPT. Furthermore, a resilience governance strategy with policy regulation as the core, technological development as the priority, and collaborative construction as the foundation is proposed. In the future, the application and development of ChatGPT need to be further integrated and constructed in the resilience governance paradigm.#br#Overall, although it is difficult to avoid or even eliminate the increasing uncertainties brought about by artificial intelligence technologies such as ChatGPT, it is certain that humans and technologies are always in an interdependent and constantly deductive relationship, and the two will continue to couple, generate, and reshape in their joint development. In response to the current trend of "practice surpasses theory" and "industry surpasses academia" in ChatGPT, the relevant research still lacks a systematic governance paradigm with conceptual and practical tension. Thereby, a comprehensive deconstruction and systematic construction of ChatGPT applications have been carried out through the resilience perspective and its research paradigm, which helps artificial intelligence applications such as ChatGPT release social value with more resilience tension to continuously evolve and improve, then promoting artificial intelligence towards a sustainable and progressive development.#br#
|
|
Received: 26 May 2023
|
|
|
|
|
|
[1] 姚叶. 人工智能算法专利的技术、理论、问题与中国之应对[J]. 科技进步与对策, 2022,39(16): 134-141.
[2] 黄荣, 吕尚彬. ChatGPT:本体、影响及趋势[J]. 当代传播, 2023,39(2): 33-38,44.
[3] 赵精武, 王鑫, 李大伟,等. ChatGPT:挑战、发展与治理[J]. 北京航空航天大学学报(社会科学版), 2023,36(2): 188-192.
[4] 陈永伟. 超越ChatGPT:生成式AI的机遇、风险与挑战[J]. 山东大学学报(哲学社会科学版), 2023,73(3): 127-143.
[5] 杨望, 徐慧琳, 王钰淇. 构建GPT大模型的经济循环体系:理论框架与发展路径[J/OL].新疆师范大学学报(哲学社会科学版):1-12 [2023-10-25]. DOI:10.14100/j.cnki.65-1039/g4.20230701.003.
[6] 黄璜. 赋物以“智”,赋治以“慧”:简论ChatGPT与数字政府建设[J].学海, 2023,34(2): 35-40.
[7] 陈智, 陈昊. 可供性视角下ChatGPT的赋能效用及其风险透视[J]. 情报杂志, 2023,42(7): 131-139.
[8] 于水, 范德志. 新一代人工智能ChatGPT的价值挑战及其包容性治理[J]. 海南大学学报(人文社会科学版), 2023,41(5): 82-90.
[9] 许源源,陈智.新一代人工智能技术社会化应用的脆弱性风险及其韧性治理研究——以ChatGPT为例[J].电子政务, 2023,20(9): 39-49.
[10] 李猛. 人工智能时代的社会公正风险:何种社会？哪些风险[J]. 治理研究, 2023, 39(3): 118-129,160.
[11] ZHANG X J, ZHANG Z G. How do smart villages become a way to achieve sustainable development in rural areas? smart village planning and practices in China[J]. Sustainability, 2020,12(24): 1-20.
[12] 唐任伍, 郭文娟. 乡村振兴演进韧性及其内在治理逻辑[J]. 改革, 2018,34(8): 64-72.
[13] 张震, 唐文浩. 韧性治理共同体:面向突发公共风险的乡村治理逻辑[J]. 南京社会科学, 2022,33(10): 54-62.
[14] 汪辉, 徐蕴雪, 卢思琪, 等. 恢复力、弹性或韧性——社会-生态系统及其相关研究领域中“Resilience”一词翻译之辨析[J]. 国际城市规划, 2017,32(4): 29-39.
[15] 中青在线.《中国人工智能大模型地图研究报告》发布,发展态势可圈可点[EB/OL].(2023-05-28)[2023-07-11]. https://news.cyol.com/gb/articles/2023-05/28/content_pQ3l4YfYBE.html.
[16] 方兴东, 钟祥铭. ChatGPT革命的理性研判与中国对策——如何辨析ChatGPT的颠覆性变革逻辑和未来趋势[J]. 西北师大学报(社会科学版), 2023,60(4): 23-36.
[17] 姚晓丹. ChatGPT的应用:从日常生活到专业领域[N]. 中国社会科学报, 2023-02-20(3).
[18] GUNDERSON L H, HOLLING C S. Panarchy: understanding transformations in human and natural systems[M]. Washington D.C: Island Press, 2002: 25-62.
[19] CHRIS S, RICHARD V N. What ChatGPT and generative AI mean for science[J]. Nature, 2023,614(7947): 214-216.
[20] NOY S, ZHANG W. Experimental evidence on the productivity effects of generative artificial intelligence[J]. Science, 2023,381(6654): 187-192.
[21] Nature. How researchers are using ChatGPT[EB/OL]. (2023-02-28)[2023-07-13]. https://www.nature.com/articles/d41586-023-00603-2.
[22] DeepTech深科技. Science封面论文:ChatGPT缩小了人们在写作能力上的差距[EB/OL]. (2023-07-14)[2023-07-18].https://baijiahao.baidu.com/s?id=1771391062176337655&wfr=spider&for=pc.
[23] OpenAI. GPT-4 Technical Report[EB/OL]. (2023-03-14)[2023-07-01]. https://arxiv.org/pdf/2303.08774v3.pdf.
[24] 刘鲁宁, 刘勉. 辩证看待ChatGPT影响智能社会治理的效用[N]. 中国社会科学报, 2023-03-06(6).
[25] Future of Life Institute. Pause giant AI experiments: an open letter[EB/OL]. (2023-03-22)[2023-06-18]. https://futureoflife.org/open-letter/pause-giant-ai-experiments/.
[26] 刘丹鹤, 孙嘉悦. 人工智能规制政策制定的风险与治理[J]. 人文杂志, 2023,67(2): 121-128.
[27] 网易网. ChatGPT引爆澳洲裁员潮！大厂华人直接丢饭碗！10大最容易被替代职业发布[EB/OL]. (2023-03-03)[2023-06-19]. https://www.163.com/dy/article/HUTOQAJH0534I43Y.html#post_comment_area.
[28] OpenAI. ChatGPT outage: here's what happened[EB/OL]. (2023-03-24)[2023-07-19]. https://openai.com/blog/march-20-chatgpt-outage.
[29] 搜狐网. ChatGPT网站访问量连降三月 学校开学或助其回升[EB/OL]. (2023-09-08)[2023-09-08]. https://www.sohu.com/a/718692190_362225.
[30] 澎湃网. 《自然》杂志:ChatGPT用于研究的五个重点问题[EB/OL]. (2023-02-07)[2023-06-23]. https://m.thepaper.cn/baijiahao_21840479.
[31] 中央网络安全和信息化委员会办公室. 生成式人工智能服务管理暂行办法[EB/OL]. (2023-07-13)[2023-07-25]. http://www.cac.gov.cn/2023-07/13/c_1690898327029 107.htm.
[32] 王英明, 汪怀君. 象脑与共情:ChatGPT的伦理风险发微与纠偏[J]. 理论导刊, 2023,45(7): 108-113.
[33] 张宇轩. 人机对话中个人信息的“设计保护”——以ChatGPT模型为切入点[J]. 图书馆论坛, 2023,43(8): 77-87.
[34] 龙柯宇. 生成式人工智能应用失范的法律规制研究——以ChatGPT和社交机器人为视角[J]. 东方法学, 2023,16(4):44-55.
|
|
|
|
