This article is by Sarah Chea and read by an artificial voice.

AI has become an essential tool - indispensable not only in biomanufacturing but increasingly in the development of pharmaceuticals, say scientists who believe it may ultimately help extend human life.
"AI is being woven into nearly every facet of the bioprocess - from enzymes and microbes to fermentation and purification," said Professor Lee Sang-yup, vice president for research at KAIST who studies chemical and biomolecular engineering, at the 2025 Daesung Haegang Science Forum held Tuesday at the Westin Josun Hotel in central Seoul.
Lee introduced his own research on the classification of enzyme functions using AI - a comprehensive study that traces the evolution of predictive techniques from early sequence-similarity models to transformer-based large language models (LLMs).
Traditionally, enzyme functions were inferred by comparing the amino acid sequences of proteins, a method that often fell short in predicting the behavior of newly discovered enzymes. But with the help of transformer-based LLMs, they allowed scientists to treat protein sequences much like natural language, dramatically enhancing the precision with which enzyme functions can be predicted.
"Deep learning technologies, in particular, can automatically extract key features related to catalytic functions from amino acid sequences, not only improving the accuracy of predictions but also opening the door to designing entirely new enzymes with functions unseen in nature through generative AI models," Lee said.
"AI for bioengineering is no longer a choice - it is a necessity."
This year's Daesung Haegang Science Forum, hosted by Daesung Haegang Science and Culture Foundation and co-organized by Daesung Group and KAIST, was held under the theme of "Bio-Innovation in the AI Era," highlighting the convergence of AI and biotechnology to showcase the latest achievements and future directions.
Launched in 2017, the annual forum, now in its eighth year, initially had "Microbes" in its name but added "Science," since last year, expanding the scope to include compelling issues in science and technology.

Philip M. Kim, a molecular genetics professor at the University of Toronto, also presented his research on using AI to design proteins.
Even subtle changes in human proteins can lead to intractable diseases such as Alzheimer's and Parkinson's, making AI-driven protein design a crucial key to next-generation cure development.
Proteins are long chains of amino acids linked by peptide bonds that fold into intricate three-dimensional structures. Depending on how these amino acids fold, a protein may become an enzyme, an antibody, or something entirely different, meaning that even the slightest error in this folding process can trigger disease.
"AI goes from coding of structure, and basically every AI method today is a sum version of this graph kind of coding of structure, and trains a neural network to do protein signal design," Kim said.
"We'll likely see the first AI-designed antibody enter clinical trials within the next year or two."
Other speakers at the event included Hwang Hee, CEO of Kakao Healthcare, who gave a speech about incorporating big data and AI to realize digital healthcare in the aging society, while Professor Zhang Byoung-tak, head of the AI Institute at Seoul National University, shared how AI can be utilized in the means of boosting the bio industry.
The Tuesday event was also attended by Daesung Group Chairman Kim Young-hoon, lawmaker Kim Sang-hoon from the People Power Party, Lee Kun-woo, president of Daegu Gyeongbuk Institute of Science & Technology, and Kim Dong-myung, president of the Korean Society for Biotechnology and Bioengineering.
"AI, which represents both the present and the future of industry, will not only transform the trajectory of science and technology but also play a groundbreaking role in national economic development," Daesung chairman Kim said.
"Through AI, I hope ...

Ver todos los episodios