When ChatGPT stands at the entrance of the university, students | knowledge | university

College students use ChatGPT to do homework and papers, while graduates use ChatGPT to prepare resumes and interviews... The various phenomena caused by ChatGPT in the field of higher education have received great attention. With the rapid development of artificial intelligence, what changes will occur in the way humans acquire knowledge? What changes will occur in the knowledge system and cognitive style? Is the existing education ready to respond?

The reporter invited Professor Zhao Bin from the School of Life Sciences at Fudan University, Professor Gu Xiaoqing from the Department of Education at East China Normal University, and Dean Gao Qiqi from the School of Political Science at East China University of Political Science and Law to jointly discuss the challenges faced by higher education in the AI era.

Introducing ChatGPT in teaching

Liberation Weekend: This semester, Teacher Zhao Bin attempted to introduce ChatGPT into teaching. Can you introduce the specific situation?

Zhao Bin: I introduced ChatGPT in the course "Ecology: The Economics of Managing Nature", mainly using a teacher guided+self-directed learning model.

I carefully selected some reports from Nature, Science, and the Proceedings of the National Academy of Sciences to determine several themes for students to read. After reading these cutting-edge articles, students inevitably have some questions and need to search and explore more information. At this point, they can ask ChatGPT questions.

In order to make ChatGPT accessible to all students, I created a WeChat group called "Ecology | Asking Good Questions to AI" and set up a robot called "Abao" in it. This robot is connected to ChatGPT, and as long as you @ it, you can ask ChatGPT questions. Of course, students also need to rely on their knowledge accumulation and judgment to identify the answers provided by ChatGPT.

Professor Zhao Bin from the School of Life Sciences, Fudan University

Liberation Weekend: Why did you attempt to ban the use of ChatGPT when multiple universities and research institutions around the world issued bans?

Zhao Bin: I am a user of ChatGPT and New Bing myself. In my conversations with them, I feel the great convenience that human-computer interaction brings to learning. This not only greatly improves our efficiency in obtaining information, but may also provide us with a more effective learning process. Attempting to introduce ChatGPT, I hope to incorporate problem driven teaching methods throughout the course to enhance students' ability for self-directed learning.

Furthermore, with the continuous development of artificial intelligence, we must face its impact and challenges on higher education. If higher education cannot always exclude artificial intelligence from teaching, then we will inevitably transform it into a teaching tool. My course experience may serve as an exploration and attempt to prepare for the upcoming educational transformation.

Liberation Weekend: So far, how is the effect?

Zhao Bin: From current observations, the vast majority of students have performed very well, especially middle level students who have clearly become more capable of asking questions and have begun to understand the true essence of critical spirit.

Students gradually realize the importance of asking questions, as only appropriate questions can provide them with ideal answers. In fact, the process of thinking about a problem is the process of establishing a connection with the external knowledge world, and ChatGPT has a strong ability to connect. If you are unable to obtain an answer in it, you must reflect on yourself and reorganize your questioning approach.

For example, when exploring complex issues such as the feasibility of reviving extinct dodo birds and ethical controversies, students need to break down the questions and gradually pursue them to obtain answers, ultimately obtaining specific evaluations and suggestions.

Liberation Weekend: How did you assess students at the end of the semester?

Zhao Bin: For the assessment of students, I no longer focus on the conclusions they draw in the end, but more on how the questions they raise are followed up again after receiving a response from ChatGPT.

At present, I mainly evaluate through watching student WeChat group recordings, which also eliminates some pain points in previous classroom teaching. For example, in group discussions, some students may not participate or may not take it seriously even if they participate; The efficiency of centralized discussion is not high, and each person's contribution to the group's results varies; The process of discussing and thinking cannot be shared among different groups. These issues were naturally resolved after the introduction of ChatGPT and new evaluation mechanisms.

Changing educational evaluation requires two major supports

Liberation Weekend: This kind of repeated exploration, questioning, and communication between humans and machines reminds us of Socratic dialogue.

Zhao Bin: The process of human-computer interaction is actually a process of mutual learning and promotion. In fact, human thinking about many problems does not have a standard answer, but humans can gain inspiration and refinement through communication with others or machines. It is entirely feasible to view ChatGPT as a knowledgeable friend or teacher.

Gao Qiqi: French philosopher Michel Montaigne believed that "the most effective and natural way to train thought is to talk to people." Dialogue plays a very important role in education, and Socrates referred to it as "midwifery." Plato's "Dialogues," Confucius' "Analects," Wang Yangming's "Transmission and Learning," and others all convey the fruitful results of human thinking in similar forms.

Unlike traditional dialogue in education, ChatGPT can not only be used for human-machine dialogue in a single scenario, but also complete multiple scenarios and rounds of dialogue. This new paradigm of "human like" interactive learning through natural language enables ChatGPT to quickly and continuously learn professional knowledge in various fields through multiple rounds of interaction, and has the potential to reach a professional level. Perhaps in the near future, humanity will usher in a revolution of dialogue in education.

Gao Qiqi, Dean of the School of Political Science at East China University of Political Science and Law

Liberation Weekend: If we continue to try to apply artificial intelligence such as ChatGPT to the education field, what risks should we be vigilant of?

Gao Qiqi: From a micro perspective, students may only receive unidirectional knowledge feed when using ChatGPT to complete homework. Their thinking, integration, and construction processes are all replaced by machines, making them easily immersed in the answers generated by ChatGPT, leading to a shallow cognitive structure and homogenization of cognitive thinking.

From a macro perspective, there is a huge explanatory black box behind general models such as ChatGPT, making it difficult for people to control the textual knowledge generated within them. Once the ChatGPT corpus is "ethically contaminated" or "data poisoned," it will pose a huge threat to adolescents whose worldviews, outlooks on life, and values have not yet been fully established.

In addition, good education requires the construction of real and effective social connections, but in the combination of universal models and education, the real interaction between learners is weakened. The loss of this learning community may gradually affect the construction of social relationships, and emotional factors will gradually dissipate in the intimate interaction between humans and machines.

Liberation Weekend: Once artificial intelligence is introduced into the education system, does it mean that the evaluation methods and standards of education need to accelerate transformation?

Gu Xiaoqing: In 2020, China issued the "Overall Plan for Deepening the Reform of Education Evaluation in the New Era", which mentioned "reversing the unscientific orientation of education evaluation and resolutely overcoming the stubborn diseases of only scores, only enrollment, only diplomas, only papers, and only hats", indicating that the country attaches great importance to evaluation reform.

However, due to the complex social functions undertaken by education and the heavy social expectations carried by the education system, it poses certain difficulties in the specific formulation and implementation of evaluation standards. At present, China has made many explorations and experiences in this field, but due to factors in the structure of the education system, it has not yet been fully implemented.

In my opinion, changing educational evaluation standards requires two major supports. On the one hand, there is a strong social consensus that only when people are fully aware of the changes in cognitive and production methods in the era of artificial intelligence, can they truly support evaluation reform and reduce resistance. On the other hand, educational informatization is an endogenous variable in the systemic transformation of education. Only by ensuring the informatization of the teaching process and teaching evaluation process can people truly believe in the objectivity and fairness of the new evaluation standards.

Professor Gu Xiaoqing from the Education Department of East China Normal University

Zhao Bin: Measuring the progress of a learner is indeed the biggest challenge that every teacher faces. In the past, teachers could judge students' learning outcomes through exams, papers, and other methods. But what about in the future?

I think with the help of intelligent devices, teachers can also explore more. For example, the artificial intelligence supported mathematics learning platform MATHia developed by Carnegie Learning can already use machine learning algorithms and predictive models to evaluate students' knowledge and skill levels. Teachers can use "knowledge tracking" to understand students' understanding of different concepts, observe their problem-solving methods through "model tracking", and adjust software to support individual students' thinking processes. This means that teachers can not only complete personalized learning evaluations, but also provide personalized learning guidance through personalized evaluations.

A new round of cognitive revolution is taking place

Liberation Weekend: Introducing ChatGPT in the classroom is an experiment, which also indicates that the application scenarios of artificial intelligence are very close to us.

Gao Qiqi: From the development history of artificial intelligence, there have been three waves: the 1950s to 1970s was the first wave, the 1980s to 2000 was the second wave, and 2006 to today is the third wave. The victory of AlphaGo over Li Shishi in 2016 was a small climax in the third wave of artificial intelligence. The birth of ChatGPT can be seen as the second small climax, and the changes it triggered have just begun, and its effects will gradually be released.

Unlike independent iconic events like AlphaGo, ChatGPT did not attract much attention when it was first released in November last year. I attended some seminars at the time, and industry insiders were not particularly surprised by its birth. It was not until January this year, when its user base exceeded 100 million, that ChatGPT gained the attention of major mainstream media worldwide and quickly sparked public discussion. When everyone can face artificial intelligence directly, the impact and shock, as well as the doubts it brings, are enormous.

Zhao Bin: The ChatGPT technology itself has undergone a long period of discussion, and its core technological innovation has already occurred. Therefore, what surprises everyone about ChatGPT is not the technology itself, but rather its ability to become a language for communication with everyone, and everyone can turn it into a tool.

Once upon a time, the cost of communicating with computers was high, and we needed to learn various programming languages, which required constant iteration. But now, we can directly communicate with machines using natural language without having to compromise on machine understanding to learn computer language, which marks the true arrival of artificial intelligence by everyone's side.

Liberation Weekend: What changes have the development of artificial intelligence brought to people's way of understanding the world? What challenges will this pose to universities?

Gao Qiqi: Currently, the "knowledge revolution" triggered by the universal big model represented by ChatGPT is flourishing, driving knowledge change, efficiency change, and power change, becoming a new way for people to understand, understand, and transform the world. It can be said that it is a super tool.

Firstly, as a proxy technology, it can arm people and enable learners to quickly acquire relevant knowledge and information, making it an efficient tool for writing assistance. Secondly, as a type of "prosthetic limb", it can narrow the gap between knowledge disadvantaged individuals and others, providing a new possibility for "knowledge equality" for families with weaker cultural capital and economically disadvantaged learners. Thirdly, ChatGPT will also generate a new way of content production and acquisition. It enables humanity to achieve a new leap from producing knowledge from experience to producing knowledge from data, and comprehensively surpasses the old content production from multiple dimensions such as content production speed, production quantity, and creativity.

Gu Xiaoqing: In the era of artificial intelligence, the updating and growth rate of knowledge is much faster than before, and the production mode of knowledge is moving towards semi automation or automation, thus forming a form of "soft knowledge". Knowledge is not only statically recognized, but also put into dynamic application.

In addition, the subject of knowledge creation has also undergone changes. In the past, universities and research institutions were responsible for the main part of knowledge creation, but now, the main body of knowledge creation has gradually shifted to research communities that are no longer limited to academia, and they are often closely connected to real production.

For example, Huawei's research team can directly form a closed loop between their knowledge creation and social production, and their research results can be quickly transformed into products and pushed to the market. Market feedback and demand, in turn, continue to drive the deepening of research and development work, forming a fast knowledge iteration and a short knowledge production loop. For universities, if it takes 10 years for cutting-edge knowledge to enter the classroom, then students' cognition is completely lagging behind.

Liberation Weekend: Due to changes in knowledge production and innovation methods, the relationship between education and industry has also changed?

Gao Qiqi: One is the change in working methods, where a large amount of work that was previously done by humans will be done by machines. This is a typical change that everyone can see. In addition, the organizational structure may undergo significant changes, with large companies becoming small companies and small companies becoming individual companies, further amplifying individual creativity.

At present, this trend has become apparent. Many technology companies abroad have very few employees, and the artificial intelligence illustration company MidJourney has an annual revenue of over 100 million US dollars, while it only has 11 employees.

Therefore, in this process, the degree of cooperation between individuals and artificial intelligence will become a crucial factor in improving productivity. Lifelong learning, interdisciplinary learning, and diversified learning will become an inevitable situation for future humanity. And society needs to provide these creative individuals with digital basic services, such as sufficient computing power, a secure digital environment, and other guarantees.

Liberation Weekend: There are reports abroad that the number of universities in the future will sharply decline, and technology giants will replace universities to cultivate the talents they need. Is it possible for this to happen?

Gu Xiaoqing: I don't think universities will be completely replaced. They will definitely still have their functions and value to undertake in the future. But various sectors of society may indeed play an increasingly important role in talent cultivation.

In fact, education was originally intended to provide training for the talents needed for industrial labor, but later the division of labor became increasingly refined, forming school education. When various sectors of society can provide more sources and conditions for talent cultivation, then breaking the single education and training system and forming a knowledge flow and creation throughout society may also be called another form of "university".

Human machine collaboration has become an inevitable trend in development

Liberation Weekend: Will the goals of education also change under this tremendous transformation?

Gu Xiaoqing: In 2016, a report from the World Economic Forum predicted that humanity should possess key abilities such as complex problem-solving, critical thinking, and creativity in addressing the challenges brought about by the Fourth Industrial Revolution. In this context, future education needs to deeply recognize the key abilities that distinguish humans from intelligent machines, shift towards promoting student growth as the center, and help students have the knowledge and abilities to adapt to future society.

The essence of education has not changed, it still promotes comprehensive human development, and the goal of education remains consistent, still cultivating innovative talents. However, innovation is not taught, it corresponds not to the inheritance of knowledge, but to the creation of knowledge and the ability to ask questions. Therefore, the exploration and cultivation of innovation require a systematic layout of education.

Liberation Weekend: What measures are currently in place in various countries to promote the cultivation of innovative talents?

Gao Qiqi: In the field of artificial intelligence, whether it is the hardware field such as chips or the application of artificial intelligence scenarios, the United States has an undeniable first mover advantage. In addition, the main R&D personnel of AlphaGo's development company DeepMind come from the UK, and Canada has been deeply involved in the field of deep learning, giving birth to many fundamental theories and algorithms... These achievements are closely related to the country's cultivation of artificial intelligence talents. For example, the United States advocates for AI education to cover all stages of education and encourages the workforce to undergo retraining; The UK advocates the development of multi-level AI skilled talents, focusing not only on high-level AI research and development experts, but also on national STEM education and data skills training.

Gu Xiaoqing: In China, curriculum reform has already reflected the transformation of talent cultivation methods. In the new curriculum plan, there is no longer an emphasis on mastering knowledge and passing exams, but more emphasis on cultivating students' core competencies.

In recent years, quality education, ideological and political education in curriculum, and strengthening and improving physical and aesthetic education in schools in the new era have all been included in national education documents. Developing skills and competencies through project-based and comprehensive practical approaches, such as interdisciplinary courses ranging from STEM to STEAM, and 21st century core competencies ranging from 4C to 5C, all indicate that China is already preparing accordingly.

In terms of interdisciplinary education, the new national curriculum standards have explicitly stipulated that we should allocate 10% of the class hours for each subject to carry out interdisciplinary education. In the future, in multi type, multi-level, and interrelated courses, students can try to find their passions and further explore and delve into their areas of interest.

In addition, in terms of industry education integration, China continues to support industry university research cooperation in education. Under the collaboration of government, enterprises, and universities, research institutions such as Huawei, Baidu, and Alibaba have shown outstanding performance in industry university research cooperation in education. Huawei's "fertile ground" AI developer enablement plan and Baidu's Cloud Intelligence College all connect AI research, entrepreneurship, investment, talent cultivation, and more.

Liberation Weekend: How will humans and artificial intelligence interact in the future society?

Gu Xiaoqing: The book "Singularity Approaches" once proposed that human-machine collaboration will become an inevitable development of human society, because the explosive growth of technology and knowledge will surpass the evolution of human physiological functions.

In fact, the concept of human-machine collaboration is not far away and may not necessarily refer to brain computer interfaces. Human machine collaboration is actually about people using technology to interact with the world, understand the world, and transform the world. If viewed from a historical perspective, today's human life is actually a "human-machine collaboration", and we can no longer do without electronic devices such as mobile phones and computers. The relationship between humans and artificial intelligence is not about who replaces who, but about both parties working together with each other's support to promote the creation of knowledge and value, which is also the greatest significance of human-machine collaboration.

Zhao Bin: The advantages of artificial intelligence lie in its speed, accuracy, and consistency. But its creativity, innovation, critical thinking, problem-solving, socialization, leadership skills, empathy, collaboration, and communication are still weak. When it comes to routine or structured tasks, automation can be completed quickly, but it still outperforms humans in terms of "soft power".

Liberation Weekend: What will the future educational venue look like?

Zhao Bin: Firstly, the classroom will become more intelligent. Creating value for teachers and classrooms through systems utilizing sensors and IoT technology; Secondly, teaching equipment will become more intelligent and specialized; Finally, the traditional structure of the class system will be broken, and personalized learning experiences will be provided through artificial intelligence exercises tailored to each student's interests and skill levels.

Gao Qiqi: General AI such as ChatGPT will become one of the most important infrastructures in the future. Enterprises, universities, social groups, etc. may all access this model, which will bring huge systematic effects in all aspects and greatly promote the iterative updating of their respective fields. On this basis, the knowledge production and transformation of universities will reach a higher level, and students will gradually find ways to dance with machines in the interaction with AI.

Gu Xiaoqing: I hope that future schools can truly break through the "wall" and establish a stable relationship between schools and other places with educational functions and knowledge production capabilities. For example, public educational venues such as science and technology museums, various research institutions, and major industrial innovation research units can truly play the core functions of schools.

In the future, school education must teach students how to collaborate with artificial intelligence technology, nurture learners' ability to learn, attach great importance to the cultivation of students' ability to analyze knowledge, summon learners to learn, and promote learners to update and create knowledge in human-computer interaction.