“Improving the performance of artificial intelligence will not be just the job of skilled developers. Any chatbot user can contribute to eliminating AI illusions and improving accuracy.”
Professor Hyungnam Moon of Sookmyung Women's University, who defines artificial intelligence (AI) and ESG (environment, responsibility, transparent management) as 'mega trends' and emphasizes them as recent growth engines in Korea, said, “Now we want the slogan 'AI first'.” opened.
Professor Hyungnam Moon is a man-made intelligence (AI) convergence business expert and serves because the president of the Korea AI Education Association and the director of the ESG Metaverse Development Institute. He can be an 'advanced technology convergence expert' and futurist who regularly appears in government policy discussions and lectures.
Particularly, the code ‘AI, ESG transformation’ has been emphasized since last yr. Related special lectures are also repeatedly conducted.
AI and ESG are areas by which many firms and governments are putting their efforts, to the extent that they were treated as major keywords on the Davos Forum, which ended last week, and the CES held early this yr. Nevertheless, AI is well known because the domain of huge tech or developers who develop models, and ESG is perceived as too broad and an abstract concept. Additionally it is true that it will not be clear at first glance what sort of relationship these two have.
Professor Moon identified that this problem was “as a result of insufficient or incorrect explanation of fundamental concepts.” He said, “To start with, within the case of AI, users may also be the protagonists in training AI.”
Typically, AI model training requires the use of specialised methods comparable to supervised learning or unsupervised learning. As well as, methods comparable to ‘Reinforcement Learning Based on Human Feedback (RLHF)’, which improves accuracy through feedback on the answers provided by AI, are well often called the strategy used to coach ChatGPT.
The purpose that Professor Moon points out is that the conversations we now have with the AI chatbot are stored back within the model. In other words, AI models expand their breadth of information through conversations with humans.
The strategy he presents is easy. “We’d like to offer feedback comparable to ‘correct’ or ‘unsuitable’ on the outcomes produced by the chatbot.” The reason is, “As feedback from hundreds or tens of hundreds of individuals accumulates, the accuracy of the model is sure to extend.”
Nevertheless, to this point, most individuals have stopped in need of calling the chatbot ‘silly’ if its answer is unsuitable. “With the intention to strengthen the competitiveness of domestic LLM, it is vital to have users actively use and re-learn the information they need,” he said.
He also said, “In this fashion, firms can actively use it for marketing.” Firms where AI cannot provide proper answers as a result of lack or lack of relevant data to learn from can inject relevant knowledge into chatbots.
This relies on a fine-tuning method called ‘Zero-shot’ or ‘Few-shot’ running. This can be a method that enables the model to derive answers to related questions by inputting just some data when it has barely learned the relevant data.
He explained that this will be proven through actual experiments. In case you enter your name and what you must tell the chatbot, in some unspecified time in the future the chatbot will ask about you and print out the knowledge you entered.
“In truth, this method made many AI experts nod their heads.” It’s a 'change in pondering' by which users train models, and he explained that this was possible due to his long experience.
He earned a doctorate in business administration from Sungkyunkwan University and accomplished a doctorate in engineering at KAIST and a doctorate in North Korean studies on the University of North Korean Studies. He later worked as a researcher and reporter on the East-West Economic Research Institute, and is currently conducting research and lecturing as a professor on the Graduate School of Management at Sookmyung Women's University. Particularly, he said that he has been accumulating his AI knowledge by reading original books since he was a graduate student at Korea University in 1985.
After working as a management and ESG expert, in 2022, OpenAI announced ChatGPT and got here up with the thought of fusing AI with ESG. Currently, we’re specializing in solving social and economic problems from the attitude of applying AI technology to society and businesses.
Particularly, he said, “In Korea, a serious misunderstanding arose from simply explaining ESG by way of environment, society, and governance.” Consequently of research activities between Korea and Canada, it was explained that from an organization's perspective, it’s correct to interpret society as 'corporate social responsibility' and governance as 'transparent management'. “Subsequently, ESG is a technology that guarantees 3Ps: saving the planet, helping people, and prosperity,” he summarized.
Ultimately, AI is a technology that will be used to profit people and to profit people. That is in keeping with the essential spirit of ESG. The reason is that ESG also can achieve an awesome transformation through AI.
Lastly, he recently presented 'Climate Technology', which prevents global warming, and 'Blue Technology', which mimics and applies the ecosystem, as core next-generation technologies.
The reason is that beyond the 4th Industrial Revolution, the fifth Industrial Revolution recently proposed within the European Union (EU) can be centered around this. The keywords of the fifth Industrial Revolution are ▲sustainability, ▲human-centeredness, and ▲resilience.
Meanwhile, Professor Hyungnam Moon is recruiting students for the fifth and sixth AI&ESG expert special courses. It should be held on the sixth floor of the Taepyeong-ro 1-ga Construction Center on February seventeenth and twenty fourth, respectively. For more details official blogYou’ll be able to check it through .
Reporter Park Soo-bin sbin08@aitimes.com