Datawhale Forum
Source: 2023 World Artificial Intelligence Conference (WAIC)
foreword
On July 8, 2023, the " 2023 World Artificial Intelligence Conference " (WAIC) came to an end. On the morning of July 8, one of the main sub-forums of WAIC, "Youth Developer Talent Training Forum in the AIGC Era", was successfully held. The forum is co-hosted by Datawhale and Shanghai Magnolia Open Source Research Institute , co-organized by Hugging Face , and supported by Jina AI's exclusive partner. Talented young AI developers adapt to the new trends of the times, open up new areas of exploration, and shape new momentum for development.
Since its launch, the event has received extensive support from all walks of life. Top scholars, industry leaders, education experts, community celebrities and other important guests were invited to gather together and delivered wonderful speeches and in-depth discussions.
Keynote Speech
Jin Yaohui , executive director of Shanghai Magnolia Open Source Research Institute and tenured professor of Shanghai Jiaotong University, made an opening speech. He said: As young developers, you are the protagonists and promoters of the future of artificial intelligence . The spirit of continuous learning, the courage to explore and practice, and constantly challenge ourselves, apply artificial intelligence technology to benefit mankind and create a better world . AI talents must constantly learn new knowledge, master new skills, expand new horizons, cultivate new thinking, and create new value.
Subsequently, Wu Fei , Qiushi Distinguished Professor of Zhejiang University, Executive Vice President of Shanghai Institute for Advanced Study, Zhejiang University, and Director of the Institute of Artificial Intelligence of Zhejiang University , delivered a keynote speech " Thoughts on the Training of Artificial Intelligence Talents from the Knowledge Points of the 101 Program ". Professor Wu Fei said: With the continuous evolution of the computer curriculum system around the world, the knowledge points of artificial intelligence are gradually becoming clear. He took the lead and took charge of the "Introduction to Artificial Intelligence " textbook in the pilot work plan of undergraduate education and teaching reform in the field of computer science ( 101 Plan ), and designed the course according to the training goals of "thick algorithm foundation, cultivation of ethical awareness, crafting tool platform, and promotion of empowerment application" The outline is to promote the integration of science and education and the integration of production and education with the three-in-one means of " textbook construction, course sharing and platform efficiency ", and jointly build a large model in the field of artificial intelligence "wisdomBot " with the Higher Education Press, from the promotion of domestic The introduction of artificial intelligence technology system into teaching, the construction of artificial intelligence theme cases, the empowerment of virtual teaching and research rooms, and interdisciplinary reforms provide support to form a digital and intelligent teaching base capability.
Subsequently, under the expectation of many participants at the venue, Datawhale jointly released the "2023 China Artificial Intelligence Talent Learning White Paper" by Datawhale, Shanghai Magnolia Open Source and Open Research Institute, Hejing Technology, and Jiangsu "Internet + Education" Research Base . This white paper is the first domestic research focusing on the learning and education of artificial intelligence talents, and it is the first to propose an open source learning model and framework.
Gan Yuanqi, a member of Datawhale and Imperial College of Technology, shared the research results of the white paper. She said: Today we are in the era of rapid development of the artificial intelligence industry. Cultivation seriously lags behind the high-speed development of the industry, which is manifested in the contradiction between "the low employment rate of college graduates and the huge talent demand gap in the industry".
We need to explore a new model of talent training, and an open source learning model is proposed. Open Collaborative Learning is a collaborative learning model in an open environment. It emphasizes the interaction, sharing and cooperation among learners. By co-creating knowledge, Collaborative problem solving and innovative applications to achieve learning goals will become an important path for future talent training. The full version of the white paper is expected to be officially released in July-August 2023, so stay tuned.
After the launch ceremony of the white paper, Fan Jingjing, the founder of Datawhale, gave a keynote speech "Open Source Learning: A New Practice Suitable for Future Talent Cultivation" and shared the status quo of the gap between universities and industries. The origin of Datawhale is the future-oriented "self-help" of learners , Trying to build a bridge from universities to industries. Ability mismatch is the main problem at the moment, but the biggest problem is that we pay too much attention to ability, while ignoring the importance of interest and love. Datawhale's solution is to let learners help learners, and filter early community members through values. Encourage interest and love, use industry-university-research resources to provide guidance and support for learners, take learners as the center, establish a symbiotic open source learning ecology, return to the essence of education, return to interest and love.
Subsequently, Wang Wei, a tenured professor of the information hub of the Hong Kong University of Science and Technology (Guangzhou), gave a speech "Thoughts on Education and Learning in the AGI Era" and shared his thoughts on general artificial intelligence: Pre-AGI learns most human knowledge, skills, But there is also a bias, and AGI will revolutionize the economy, society, and our thinking. AGI accumulates knowledge and skills in a compressed form, optimizing the learning experience. The most important thing for learners is to maintain curiosity and adventurous spirit, to be both broad and profound, and use AGI to enhance effective ability!
After that, Xiao Han, founder and CEO of Jina AI, founder and chairman of the German-Chinese Artificial Intelligence Association, and member of the board of directors of the LF Open Source Foundation, gave a speech on the theme of "Redefining the New Trend of AI Developers" and analyzed the single-mode The current situation, the limitations of single-modal techniques and methods that are difficult to transfer, and the representation learning that allows the model to create a general representation for all modalities and the transfer learning that allows the model to first learn the basics and then fine-tune it in a specific domain. Advances in learning technologies have facilitated the rise of multimodal AI, a paradigm shift from unimodal to multimodal. He has insight into new directions under multi-modality, and analyzes prompt word-based optimization and service and model-based fine-tuning and service , in terms of R&D investment, investment, market demand, technical demand, privacy and data protection, output control and commercialization, etc. perspective and demonstrates the PromptPerfect offering in Prompt Optimization vs. Prompt as a Service.
Finally, Yang Yiyuan, a member of Datawhale, a doctoral student at Oxford University, and a Clarendon scholar at Oxford University, gave a speech "New Exploration of Learning under the Wave of AI". He analyzed the different stages of AIGC's iterative development, from AI perception & understanding of the world to AIGC generation & creation The world, from "super analyst" to "content inventor". AIGC's strong cross-field attributes bring opportunities and challenges in various fields , and shared the AI learning process of the designer and a PhD candidate from the Department of Medicine. In the open source tutorial EasyRL, from demanders & enthusiasts who study purposefully, to learners who self-study open source open course materials, to developers and organizers who actively accept feedback from other learners, to Actively connect with scholars and business experts, check for gaps and make up for gaps, sharers of published books, and finally return to update iterations & find insufficient learners, how to embrace the role change process of open source learning and self-growth, and share his understanding of open source learning: open source Learning = self -driven learning with decentralized feedback loop .
round table discussion
Next, this forum opened the prelude to the second session. With the theme of opportunities and challenges for talent growth in the AIGC era, we specially invited Hong Kong University of Science and Technology (Guangzhou) Information Hub tenured professor Wang Wei, Jiangnan University School of Education, Jiangsu "Internet Wang Zhijun, professor of +education research base, Xiao Han, founder and CEO of Jina AI, Wang Tiezhen, head of Hugging Face China, participated in the roundtable discussion, and Dean Jin Yaohui served as the special host. The round table forum focuses on what is the sustainable development path of talent cultivation, what are the different strategies and methods of different institutions (such as universities, enterprises, open source communities, etc.) in talent cultivation, how to cultivate cross-compound talents, and how to cultivate master-level leading talents Starting from their own rich experience in education, scientific research and entrepreneurship, several guests shared their unique perspectives and cutting-edge views.
The forum came to a successful conclusion. In the current rapid technological change in the world, we firmly believe that young developers and learners are the backbone to promote the development and innovation of artificial intelligence, and talents are the foundation of the development of the AI industry. This forum brings The most vivid thinking and experience of the participants, seeing different theories and practices, truly empowering the cultivation of AI talents, exploring new models, releasing the kinetic energy of all parties, exerting synergies, and helping Shanghai to accelerate the creation of a new highland for AI talents.
