Artificial intelligence and the humanities and social sciences go hand in hand

Author: LEI Huanjie (Deputy Director, Research Center for Science, Technology and Society, Chinese Academy of Social Sciences, Assistant Researcher, Institute of Philosophy)

Artificial intelligence is not only a problem in the fields of natural sciences and engineering technology, but also in the fields of humanities and social sciences. The focus on artificial intelligence in the humanities and social sciences, especially philosophy, has a history of nearly half a century in China, and has experienced changes in research focus in different periods. Tong Tianxiang, a pioneer who first proposed the "intelligent revolution theory" in the 80s of the 20th century, pointed out that "the development of high technology will inevitably lead to the intelligent revolution into the intelligent era" and "the main form of the new generation of productive forces will be the human-machine composite intelligent system". At that time, senior scholars predicted that human-machine composite intelligent systems in the era of intelligence would be applied to the fields of humanities and social sciences; Now, this is getting closer to reality. With the development of artificial intelligence and the deepening of its impact on all walks of life, the humanities and social science infrastructure will have different characteristics of "human-machine compounding". Recently, new advances in generative artificial intelligence have once again attracted the attention of all walks of life. In the face of the accelerated era of artificial intelligence, we should actively build new infrastructure in the field of humanities and social sciences, promote the high-quality application of new technologies such as generative artificial intelligence as soon as possible, and seize development opportunities. This is of great and far-reaching significance for comprehensively promoting innovation in various fields of humanities and social sciences and promoting the development of artificial intelligence for good.

A shift in focus due to generative AI

In today's era, the application of artificial intelligence has strong penetration, diffusion and subversion. For current generative artificial intelligence, some people marvel that this is no longer "alchemy", but the "spark of general artificial intelligence" has appeared. At the same time, this has also caused some people to have doubts and fears about the loss of control of technology and social disorder. For a time, there was a lot of controversy about generative artificial intelligence. To understand the essence of the problem, we also need to have a more comprehensive understanding of generative artificial intelligence and understand what it means to go in artificial intelligence. This can be grasped from the main characteristics of current generative artificial intelligence.

One is the ability to generate new content. Generative AI doesn't just simulate and predict based on existing data, but learns from that data and generates new content based on it. Therefore, generative artificial intelligence shows a certain "creativity", presenting the important feature of "intelligent generativeness".

The second is to have language processing capabilities. In the face of the problems raised by humans in natural language, the current generative artificial intelligence based on large language models has been able to give human-like answers to a certain extent. This is also a big reason why ChatGPT amazes and impresses many people. Relying on outstanding language understanding and text generation capabilities, current generative artificial intelligence has far surpassed previous chatbots and is expected to become a new generation of learning and work assistants widely used.

The third is to update the human-computer interaction ecology. The new progress of generative artificial intelligence once again shows that the breakthrough in function is not only the emergence of functions after the amount of data and model parameters reach a certain scale, but also the improvement of human feedback and human-computer interaction in the process of human feedback. As a result, researchers can more accurately grasp the facts and weigh the value.

The fourth is to transform the mode of knowledge production. Accompanied by the renewal of human-computer interaction ecology is the transformation of knowledge production methods. The automated knowledge generation model of "big model + human feedback" opened by generative artificial intelligence is a new synthesis for the intelligent future. Compared with the traditional knowledge production method, the knowledge production mode supported by generative artificial intelligence has the characteristics of fast speed, low cost and wide range, which can greatly improve the knowledge production efficiency.

Fifth, new requirements are put forward for people. Generative artificial intelligence often does not have the ability to identify whether the generated content is correct, accurate and related value judgments, and the evaluation of content still depends on human information literacy, knowledge system and critical thinking ability. Both data input (training, feeding, etc.) and data output (judgment, utilization, etc.) place different requirements on human capabilities.

The above characteristics show that generative artificial intelligence has important innovative significance. Many have been debating whether artificial intelligence will develop intelligence that resembles or even surpasses humans. New advances in generative AI may mean a shift in focus from theoretical to practical.

AI-driven research in the humanities and social sciences

The academic development trend in recent years shows that the frontier dynamics of science and technology lead and even determine the generation and migration of hot spots in humanities and social science research to a large extent. A new round of scientific and technological revolution and industrial transformation has gradually brought mankind into the era of intelligent interconnection of all things. The accelerated penetration and integration of emerging technologies and economic and social development will gradually cover the application scope of artificial intelligence from natural sciences and engineering technologies to the fields of humanities and social sciences, thus creating an AI-driven humanities and social science research. This means at least three levels of driving: the application of artificial intelligence to various fields of the humanities and social sciences; Artificial intelligence helps the innovation and development of humanities and social sciences; Artificial intelligence expands the humanities and social sciences about artificial intelligence. In short, the participation and support of artificial intelligence will help humanities and social science researchers to better understand the nature of human thinking and behavior, and explore the future fate of human civilization under the trend of deep science and technology.

AI-driven humanities and social science research points to a human-machine collaborative knowledge production method, which is conducive to liberating research productivity and unleashing innovation potential. The first is to further lower the threshold for acquiring knowledge. With the powerful information processing capabilities such as search and classification of artificial intelligence, researchers can obtain massive knowledge at any time, anywhere and anywhere. The second is to increase the richness of knowledge supply. Researchers can access more data that was previously difficult to obtain, breaking down barriers in fields and disciplines. The third is to realize the preliminary processing of content. Relying on the text processing and synthesis capabilities of artificial intelligence, the basic and repetitive labor in the knowledge production process of researchers can be reduced. The fourth is to provide optional knowledge. Artificial intelligence can provide all kinds of relevant information needed for research topics, allowing researchers to obtain multiple research options rather than a single research solution. The fifth is to assist in the integration and output of knowledge production results. AI can assist researchers in writing texts for routine tasks (e.g., literature reviews, research proposals, abstracts, course outlines, etc.). Sixth, feedback and correction of knowledge production process. The collaborative, connected relationship between humans and machines enables AI to support reinforcement learning based on human feedback.

From the perspective of the whole process of knowledge production, the significant changes brought by artificial intelligence also include knowledge management and dissemination, which will make knowledge more rapid and efficient management, dissemination and application. AI can provide more efficient and practical solutions for knowledge management. For example, tools such as knowledge graphs and data warehouses can help people process, analyze, and maintain large amounts of data to better organize and manage knowledge. Artificial intelligence can also provide a more convenient technical environment for knowledge dissemination, break the digital divide and data monopoly, promote knowledge exchanges and public dissemination, and improve knowledge openness. Some views describe the knowledge production mode in the era of artificial intelligence as an entangled "techno-social structure", with obvious space-time, self-organization, dynamic adaptability and system openness.

AI-driven humanities and social science research will also be integrated into the education and talent training system. The first is to integrate into curriculum teaching. Artificial intelligence can assist teachers in generating teaching materials such as course syllabuses, exercise assignments, and answering students' questions, thereby stimulating the update and adjustment of assessment methods and teaching goals. The second is to integrate into the cultivation of talents. When used properly, AI can help students train dialectical, critical, innovative and other thinking, form independent thinking and problem-solving skills, and cultivate talents in the humanities and social sciences in the new era. The third is to integrate into the construction of disciplines. Artificial intelligence can promote the emergence of new research methods and ideas, not only help the incubation of some emerging disciplines and interdisciplinary disciplines, but also help basic disciplines, key disciplines and unpopular disciplines to find new growth points, thereby promoting the development of disciplines and the construction of discipline systems.

At present, artificial intelligence has brought some risk shocks and ethical challenges that cannot be ignored, and has caused many criticisms and controversies. In the face of the reality that emerging technologies are constantly updated, iteratively and widely used, the problem of how to "make good use" and "make good use" of artificial intelligence is becoming more and more important. At the same time, "good use" and "good use" can also promote the better development of artificial intelligence in the future. For the humanities and social sciences, evolving AI is an excellent tool that can be an important research assistant. We should seize the opportunity, lay out the infrastructure and vigorously develop its huge application potential.

Build humanities and social science infrastructure in the era of artificial intelligence

Strengthening the construction of infrastructure such as books, literature, networks, and databases in the humanities and social sciences, and building a digital scientific research platform that is convenient, fast, and resource sharing is the proper meaning of accelerating academic development. Humanities and social science research in the era of artificial intelligence calls for infrastructure that can better promote academic innovation. To build an AI-driven comprehensive humanities and social science innovation platform, we need to pay attention to the following aspects in the specific practical strategies of the promotion process.

First of all, strengthen the top-level design, coordinate multiple forces, and fully integrate and utilize the existing digital infrastructure of humanities and social sciences. On the one hand, we must adhere to the publicity and inclusiveness, ensure that access is available online, and overcome differences in geographical space, economic level, and educational resources. On the other hand, it is necessary to grasp the overall situation and strategy, carry out overall design and deployment of the goal setting, rhythm grasp and resource investment of innovation platforms and related construction projects, coordinate multi-party collaborative construction, and grasp the technical support and guarantee, data supply and sharing, information security protection, dispute handling, financial and material and human resources support in the construction and maintenance process. Considering the characteristics of new technologies and the cycle law of construction projects, it is advisable to adopt phases and pilot methods. For example, interdisciplinary teams can be formed to carry out preliminary research on the main architecture, content, functions, mechanisms and rules of the innovation platform.

Secondly, overcome the shortcomings of the existing humanities and social science databases and create an innovative platform that actively serves researchers. The first is to better meet the general needs of researchers. It is necessary to optimize the document storage and search functions of traditional databases, so that the total number of documents is larger, the variety is richer, and the search is faster. Second, the user interface and experience are more friendly. It is necessary to bid farewell to the one-way usage mode of traditional platforms that only have search, complex functions, and scattered results, so that researchers are easier to adapt, faster to get started, and better to operate. Third, the new ecology of human-computer interaction has reached the effect of "1+1>2". The so-called new ecology of human-computer interaction does not refer to the replacement of human intelligence by machine intelligence, but to the effective combination of the respective advantages of machine intelligence and human intelligence. The fourth is to provide users with individualized and differentiated services. It is not only necessary to generate richer content in more scenarios and give more accurate responses to the needs put forward by users, but also to be able to pay attention to every researcher as a micro subject and obtain the needs and preferences of users in an all-round way. Fifth, play the role of senior research assistant. In the process of using the innovation platform, people's main role should be able to be more fully played.

Third, enhance the creativity of researchers and the innovation of research results. In the past, human beings were in an environment with a relatively lack of information, and the key to improving intelligence was how to obtain knowledge; Today, human beings are in the environment of information explosion, and the bottleneck to improve intelligence is how to judge, penetrate and apply the knowledge obtained. It can be expected that low-level research behaviors (such as searching for data, combing literature, writing abstracts, etc.) may be completed by artificial intelligence in the future, while high-level research behaviors (such as asking questions, summarizing experience, creating theories, etc.) may not be replaced in the future. This puts forward higher requirements for researchers' ability to innovate, and perhaps even innovation ability is an important endowment and characteristic that humans are difficult to replace by machines. After the popularization of innovation platforms, humanities and social science researchers may be able to produce more innovative results inspired by artificial intelligence and inspire epiphany. The quality of text content generated by AI often depends on the quality and quantity of the corpus learned. Through deep learning, the large-scale Chinese corpus can be used to enable the artificial intelligence driving the platform to master the grammar, vocabulary, semantics and more advanced academic history, research frontiers, professional dynamics, etc. of Chinese as much as possible, so as to generate high-quality Chinese texts.

Finally, establish and improve corresponding institutional mechanisms and actively strengthen ethical governance. It is necessary to attach great importance to system construction, continuously improve the institutional mechanism for multi-party participation and collaborative construction, establish a multi-department and multi-subject consultation mechanism, a sharing mechanism to break the data monopoly, and an interaction mechanism for all users to participate. It is necessary to adhere to ethics first, strengthen source governance, respond to ethical risks caused by new technologies in a timely manner, and clarify and adhere to certain ethical principles. First, adhere to people-oriented. It is necessary to ensure human subjectivity and autonomy, not excessive and blind pursuit of automation, avoid people tending to be passive and machined, recognize the essence of automatically generated content, and break the illusion that artificial intelligence can have an authoritative position in knowledge or morality. Second, adhere to public welfare first. It is necessary to face different types of users fairly and inclusively, and promote the healthy development of humanities and social sciences. Third, persist in controlling risks. It is necessary to fully estimate the risks and uncertainties in the application of artificial intelligence, advocate full-process follow-up assessment and multi-party consultation and consultation, prevent the misuse and abuse of artificial intelligence in humanities and social science research, ensure that the application in research, teaching and achievement transformation and other fields conform to academic norms and scientific research integrity, and take the road of responsible innovation. Fourth, adhere to openness and transparency. Social participation and brainstorming can be encouraged, the transparency of relevant information disclosure can be improved, and foreign cooperation, exchanges, promotion and dissemination can be done well, so as to better play the unique role of humanities and social sciences in integrating Chinese and foreign cultures and enhancing exchanges and mutual learning among civilizations.

There is great potential to build the infrastructure of humanities and social sciences in the era of artificial intelligence and develop AI-driven humanities and social science research. This is a vast and complex new world that requires multi-sectoral and multidisciplinary cooperation. This work is a systems engineering and a long-term undertaking, which requires continuous follow-up of new developments in artificial intelligence, big data and other fields, continuous absorption of new technologies, timely adjustment and improvement. In this way, we can promote the innovative development of humanities and social sciences that are more cutting-edge, contemporary and practical, and at the same time, we can promote the further development and application of artificial intelligence, so that artificial intelligence and humanities and social sciences can go hand in hand.

(This article is a phased achievement of the National Social Science Foundation Youth Project "Research on the Integration of Scientific Spirit and Humanistic Spirit under the Challenges of New Science and Technology Humanities" (22CZX021))

Source: China Social Sciences Network - China Social Sciences News