Wang Huaimin, academician of the Chinese Academy of Sciences: "According to my philosophical view, the big model is a tool, and technology is just a means. But the big model is not cast at once, it is a process that continues to synchronize with human progress. ”
Richard Guo, Chief Vice President of HKUST: "Smart people should use the intelligent machines we create to train smarter people, so that they can create smarter intelligent machines." This is a healthy cycle and human progress, and there is no reason to be pessimistic. ”
On July 8, at the Education Forum of the 2023 World Artificial Intelligence Conference, a number of scholars in the field of computer science and big data discussed the impact of artificial intelligence on educational reform and talent training with the theme of "Leading Learning Change, Intelligent Education Future".
"The big language model is the latest technological means to support intelligent education and deserves continuous attention, but we must not forget human subjectivity at any time." Wang Huaimin, academician of the Chinese Academy of Sciences and professor of the National University of Defense Technology, said in the keynote speech.
Yike Guo, Chief Vice-President of the Hong Kong University of Science and Technology and Academician of the Royal Academy of Engineering, also emphasized the fundamental impact of artificial intelligence on education, "Smart people should use the intelligent machines we create to train smarter people, so that they can create smarter intelligent machines." This is a healthy cycle and human progress, and there is no reason to be pessimistic. ”
Large models can exercise students' critical thinking
Wang Huaimin pointed out that the improvement of human education ability comes from the improvement of the means and level of human beings to stimulate and gather group intelligence. "Where does a human-acceptable high-quality model come from? It is first generated from high-quality data generated by human groups. Intelligent education can become the source of the continuous growth of large models and eventually evolve into high-quality human feedback data. ”
Wang Huaimin, academician of the Chinese Academy of Sciences and professor of the National University of Defense Technology, delivered a keynote speech at the Education Forum of the 2023 World Artificial Intelligence Conference on July 8.
Wang Huaimin also believes that when humans see the progress of large models today, they should not ignore and forget the important role played by swarm intelligence in the process of building large models. "According to my philosophical view, the big model is the tool, and the technology is just the means. But the big model is not cast at once, it is a process that continues to synchronize with human progress. ”
The big language model also has multiple implications for the development of intelligent education. On the one hand, large language models may become intelligent assistants for "good teachers". Wang Huaimin said that it is very important that good teachers can give "human judgment" based on the quality of the feedback data from large models. "Because this judgment will be fed back into the human research space again, and then form a readjustment of the large model based on human feedback data." At the next inquiry, the adjusted model has the potential to generate a new answer that satisfies and even inspires the teacher, and the optimized answer can use high-quality feedback for the teacher's teaching activities. ”
In addition, a good big language model may also become an intelligent assistant for "good students". Wang Huaimin said that when students are in the entry stage, the large model can feedback conclusions that are comparable to the student's level, and further help teachers judge the degree of students' understanding of relevant issues. "How do teachers judge a student's learning status? We usually look at students' questions, understanding and reactions to system feedback, and feedback from large models can help teachers understand students' learning status, so it is a very good tool. ”
In Wang Huaimin's view, the big model can exercise students' critical thinking and question the existing state of knowledge. "As students ask deep-seated questions, large models can help teachers judge students' progress in learning. This data from students may become high-quality data that can be reused to train high-quality models. Wang Huaimin further pointed out from the perspective of education that the big model is the latest technology that deserves to be embraced and paid attention to. If students can question feedback on large models, this will be another effective tool for us to train students. ”
But Wang Huaimin also stressed that the big language model is the latest technological achievement that gathers the intelligence of human groups, not the only technological achievement. "The big model is the latest technology supporting smart education, and its impact and potential continue to show. We still need to pay attention to the continuous learning, growth and evolution of human-computer collaboration, and the construction of a technical system for the common progress of big models and human beings. ”
The Hong Kong University of Science and Technology has taken the lead in embracing artificial intelligence
Guo Yike said that as a world-class science and engineering university, HKUST has fully applied ChatGPT to teaching and is the first university in Hong Kong to offer ChatGPT for students and staff. "Our school established its own ChatGPT and wanted to further analyze its impact on education and thus transform our university education." For the goal of university education in the era of artificial intelligence, Guo Yike hopes that in the future, "we can really do a good job in the next generation of artificial intelligence systems, and train students to become a new generation of learning with artificial intelligence." ”
Yike Guo, Chief Vice President of the Hong Kong University of Science and Technology and Fellow of the Royal Academy of Engineering, delivered a keynote speech at the Education Forum of the 2023 World Artificial Intelligence Conference on July 8.
Guo Yike also believes that human education and machine education are polar. "The order of human education is to first cultivate common sense and curiosity in early childhood, then shape values in adolescence, then systematically learn mathematics, physics, geography, history and other subjects, accept indirect knowledge, and finally experience and self-learning in society."
But the learning process of machines is diametrically opposed to human education. "Machines first put all of human data into the system, learning models from different domains, such as medicine and physics. Going one step further, machines need to know what is right and wrong, start learning values, and get feedback from humans. Finally, the machine tried it, which is completely reverse from the steps of human education. Guo Yike believes that this is the reason why humans learn with limited energy for millions of years, while generative AI consumes a lot of energy to learn in hundreds of days." Therefore, we believe that it is necessary to better apply ChatGPT to teaching. ”
Guo Yike also shared the help of ChatGPT for students in the forum. "We started by letting ChatGPT play a variety of different roles as an assistant to help students, even teachers and research teams, gain more useful experiences. At the same time, we hope that ChatGPT, as an intelligent tutoring system, can enable students to better achieve personalized learning. ”
In addition, Guo Yike said that artificial intelligence has creative applications in future education, human beings from consumers of knowledge to co-creators, and students also need to change from passive knowledge recipients to active explorers. "When you learn, you don't know the ultimate purpose of your learning, because your purpose is to explore. And now the machine plays the role of exploration, and you, as a human being, carry out doubts about the original knowledge. Therefore, we are currently working in the direction of extended learning, which is also the most important direction. ”
Guo Yike put forward a philosophical point at the end of his speech, "In an AI-empowered world, we first need to remember humanity and put value first. Because the ultimate criterion for humans to judge whether artificial intelligence is correct is the concept of 'human', and humans and machines are different. ”