ChatGPT and Educational Transformation

ChatGPT and Educational Transformation

This Article Overview

10000

Words

Reading Time

20

Minutes

Authors | Wang Zhuli, Wu Yanru, Wang Yun

Source | Distance Education Journal, edited

The emergence of generative artificial intelligence represented by ChatGPT has stirred profound responses in the education sector, and its popularity has not waned. What does the emergence of ChatGPT signify? What new challenges does education face in the intelligent era? How should education respond to these challenges? These are urgent questions that need to be explored in the current education field and are key issues concerning the survival and development of humanity. Therefore, this article will address these three themes and explore the challenges and transformations faced by education in the intelligent era.

1. What Does the Emergence of ChatGPT Signify

1.1 Revolutionary Progress of ChatGPT

Bill Gates considers ChatGPT to be one of the two revolutionary technologies he has encountered in his life and predicts that AI-driven software will ultimately fulfill its promise of radically changing the way people teach and learn within the next 5 to 10 years (Gates, 2023).
ChatGPT has made breakthrough progress compared to previous AI. Traditional AI was designed to perform specific tasks and required predefined rules and procedures; once beyond these boundaries, it would fail. ChatGPT is based on deep learning technology, relying on self-learning, simulation, and reasoning to generate text and dialogue. It is the first to adopt a method of reinforcement learning from human feedback (Reinforcement Learning from Human Feedback, RLHF), which has the ability to improve output results based on user feedback and continuously update and learn according to user feedback (Ouyang L, et al., 2022; Shen, et al., 2023), thus coming closer to human language and habits. To some extent, it can achieve independent thinking and creative expression, which makes it exhibit greater adaptability and applicability than traditional AI.
General artificial intelligence (Artificial General Intelligence, AGI) refers to intelligent machines that can think like humans and engage in multiple purposes. As mainstream AI research gradually moves towards intelligence in specific fields (such as machine vision, speech input, etc.), to differentiate them, the term “General” was added to Artificial Intelligence, thus opening up the field of general artificial intelligence (Baidu Encyclopedia, 2023). As a natural language processing tool driven by artificial intelligence technology, ChatGPT not only can engage in dialogue by understanding and learning human language, interacting based on the context of the conversation, truly communicating like humans and answering various questions; it can also perform multiple tasks such as writing emails, video scripts, copywriting, translation, coding, and writing papers, making significant progress towards the goal of general artificial intelligence. However, it has not yet reached the “universality” of the human brain and cannot engage in more work like humans, so it cannot yet be defined as general artificial intelligence (see Table 1).
Table 1 Similarities and Differences Between Human Brain Intelligence and ChatGPT Intelligence
ChatGPT and Educational Transformation
From Table 1, it can be seen that so far, the only intelligence that ChatGPT can match with the human brain is thinking, and in a strict sense, ChatGPT’s thinking is fundamentally different from human thinking. ChatGPT generates responses based on existing data and programs, rather than true thinking and judgment. Human thinking is a process characterized by subjectivity, free will, emotion, and perception. It often originates from human perception of the external world and itself and rises to rational understanding through forms of thinking such as abstraction and generalization based on perceptual cognition. ChatGPT does not possess these characteristics; it cannot perceive external things but searches for matching answers by processing and analyzing input natural language. The technology it uses is based on machine learning and natural language processing algorithms, establishing models by processing large amounts of training data to provide the best possible answers in given situations. Therefore, even if its responses appear similar to human thinking results, it cannot be said that its thinking is the same as human thinking. Although humans sometimes also search for existing answers in memory to match problems, human thinking is clearly much more complex and flexible than this.
Due to ChatGPT’s lack of real-time perception of the external world and abstract thinking ability, it still has a significant gap compared to the human brain in dealing with open-ended questions and reasoning in unstable scenarios. Ultimately, ChatGPT is merely a computer program composed of algorithms and mathematical models, and its working principle is vastly different from the functions, structure, and evolutionary history of the human brain. For example, humans can discover new things and phenomena from practice, name things or phenomena, and then abstract, generalize, analyze, and reason about the essence and laws of things or phenomena to produce new knowledge. So far, ChatGPT can only re-collect, organize, process, and reassemble existing knowledge and data to match the most suitable answers for specific problems based on human morals and needs.
In a sense, ChatGPT is still a “second-hand dealer of knowledge.” However, it should also be noted that this situation may change in the future. There are already intelligent machines that can perceive external information in real-time through various sensors, such as image recognition and speech recognition, and there are explorations and research using AI for emotional computing, as well as various robots that can act like humans. If generative artificial intelligence like ChatGPT is integrated with these technologies and machine systems, it may produce robots that are very close to the diversity of human intelligence, capable of engaging in various tasks like humans, not just question-and-answer dialogue. This intelligent system that can understand, reason, and interact with the physical world is called embodied intelligence and is considered one of the ultimate forms of AI (Qin Xiao, 2023). Therefore, it can be seen that ChatGPT has the potential to become a general-purpose technology (GPTs) (Fang Xingdong, et al., 2023).

1.2 The Birth of ChatGPT Marks the Official Arrival of the Intelligent Era

Whether the birth of ChatGPT signifies the official arrival of the intelligent era depends on the standards set by humanity itself. People often use the most representative production tools to represent a historical period, such as the Stone Age, Bronze Age, Iron Age, Steam Age, and Electric Age. Now, artificial intelligence has already emerged, and its importance is no less than any historical production tool, and it is certain that humanity is moving towards the intelligent era. If one must find a time point to mark the official arrival of the intelligent era, I believe the emergence of generative artificial intelligence represented by ChatGPT can be considered a landmark event. Because many previous breakthroughs in AI technology were not closely related to the daily lives of each of us, but generative artificial intelligence represented by ChatGPT may enter everyone’s daily life and have a profound impact on all aspects of society. ChatGPT attracted 100 million users in just two months, and with the continuous launch of “ChatGPT+” and similar generative AI products, it is very likely to reach a “killer application” of 1 billion users within six months to a year. Bill Gates specifically discussed the potential impact of ChatGPT on various aspects under the title “The Age of AI has begun” (Gates, 2023). The academic community seems to have yet to clearly propose what event should mark the arrival of the intelligent era. I once asked ChatGPT, and its answer was that only when machine intelligence reaches or surpasses human intelligence, possessing higher-dimensional judgment, decision-making, and autonomous action capabilities, and thus brings about a complete change to humanity, can it be said that the intelligent era has officially arrived, meaning that it must wait until true general artificial intelligence or even superintelligence arrives. This may represent the views of some experts. I agree with Fang Dongxing and others’ viewpoint: “If we change our thinking from the bottom up and change our perspective, using the widely used ‘application’ as the basis for defining ‘general,’ then ‘where many have walked, it becomes a road,’ and ‘where many have used it, it becomes general,’ rather than pre-setting a lofty goal from a ‘technology’ perspective.” (Fang Dongxing, et al., 2023)

2. What New Challenges Does Education Face in the Intelligent Era?

I previously proposed that learning in the network era faces two major challenges: information overload and knowledge fragmentation (Wang Zhuli, 2011). What challenges should learning face in the intelligent era? Do the original challenges still exist? Have new challenges emerged? Besides possibly bringing convenience to learning, does ChatGPT also bring new challenges? The challenges to learning are essentially challenges to education.
Currently, the academic community is worried about the challenges that ChatGPT brings to education, mainly focusing on false information, academic ethics, data security, digital divide, bias and discrimination, and the weakening of the teacher’s status (Wang Youmei, et al., 2023; Yang Xin, 2023; Zhou Hongyu, et al., 2023). I believe that the challenges posed by ChatGPT to education are not limited to the above aspects, but are comprehensive, involving education theory, education system, and education practice.

2.1 Challenges to Educational Theory

Educational theory is the foundation and core of the education discipline and has important guiding significance for educational reform and development. Educational theory is premised on and based on learning theory, and the two are two sides of the same coin. The emergence of generative artificial intelligence like ChatGPT forces educational theorists to answer a series of questions: (1) What is knowledge? What is useful knowledge? Has knowledge changed? What changes have occurred? (2) Why do we learn today, what do we learn, and how do we learn? (3) Why do we teach today, what do we teach, who teaches us, and how do we teach? (4) What is the main task of education in the intelligent era? What kind of talents need to be cultivated? How to cultivate such talents? These questions involve multiple levels such as the concept of knowledge, view of learning, view of teaching, view of curriculum, and view of talent.

2.2 Challenges to the Education System

The education system refers to the totality of various educational institutions and organizations established by a country or region. The education system includes educational institutions at all stages, such as preschool education, primary education, secondary education, higher education, and continuing education, as well as related educational organizations, such as educational administrative departments, educational research institutions, and educational training institutions. I divide it into two main components: the school education system and the lifelong education system. The emergence of ChatGPT also prompts us to reflect: Will schools still exist in the future? Will the education system still be the same as now? Should the status of lifelong education be elevated? How to elevate it? To what extent? What is the relationship between lifelong education and school education? and other questions.

2.3 Challenges to Educational Practice

Educational practice refers to the educational and teaching activities completed collectively by all workers and learners in the education system under the guidance of educational theory. The challenges faced by educational practice are the greatest and the most difficult to achieve in one go. This is a challenge directly faced by all educational institutions and educators, including how to transform advanced educational concepts into actual educational actions? How to achieve transformation in the school system and educational ecology? How to realize the transformation of educational teaching models and methods? How to update educational content? How to reform the educational evaluation system? How to enhance teachers’ educational and teaching capabilities? How to cultivate students’ learning abilities centered on digital literacy? and other questions.

3. How Should Education Respond to Challenges in the Intelligent Era?

Some scholars assert that artificial intelligence will render the advantages of traditional education obsolete (Qian Yingyi, 2023). However, education is a major national affair, and all aspects involved are very complex, including politics, economy, culture, and many other fields, requiring joint efforts from all sides to achieve reform. Educational reform is a long-term process that requires time to gradually promote, significant investment and effort, and the promotion of educational system reform, the updating of educational ideas, the allocation of educational resources, and other aspects of work. Therefore, the speed of educational development is often slower than the speed of technological progress, and this has become increasingly evident in the network and intelligent eras. However, education reform cannot be delayed; it is imperative. I intend to explore the content and direction of educational reform from the perspective of education itself, focusing on the following aspects.

3.1 Comprehensive Reflection and Upgrade of Educational Theory

In the intelligent era, new learning technologies and environments require a comprehensive reflection and upgrade of the basic theories of education and teaching. Below, I will analyze this from the perspectives of the concept of knowledge, view of learning, view of teaching, view of curriculum, and view of talent.
1. New Concept of Knowledge
1.1 New Concept of Knowledge in the Network Era
The initial observation of knowledge changes in the network era comes from the founders of connectivism learning theory, George Siemens and Stephen Downes (George Siemens & Stephen Downes). Siemens et al. believe that knowledge in the network era has transformed from a static hierarchical structure into a dynamic network ecology, proposing new concepts such as knowledge flow, soft knowledge and hard knowledge, and connective knowledge (Siemens, 2009; Downes, 2008). It has gradually diverged into two new concepts of knowledge:One is the new concept of knowledge oriented towards the intelligent era that I proposed in 2017, which gradually developed into the reconstructionist concept of knowledge (Wang Zhuli, 2017a; Wang Zhuli, et al., 2022). Its most significant feature is a further deepening and expansion of the distinction between soft and hard knowledge, believing that a new type of knowledge called soft knowledge has emerged in the network era, which is increasingly important today, while the importance of traditional hard knowledge is declining. The other is the knowledge concept of

Leave a Comment