There is no future unless we integrate bio, medicine, engineering, and artificial intelligence (AI).
On the 26th at the Grand InterContinental Parnas in Seoul, Research Vice President Lee Sang-yeop of the Korea Advanced Institute of Science and Technology (KAIST) took the stage as a keynote speaker at the 2026 Global Bio Conference (GBC) and said this.
Lee is a leading Korean bioengineer who has pioneered the field of system metabolic engineering by combining metabolic engineering with systems biology, synthetic biology, and evolutionary engineering. He joined KAIST as a professor in 1994 and is now a distinguished professor in the Department of Chemical and Biomolecular Engineering and the research vice president. Recently, he became the first Korean researcher to be selected as a fellow of the European Academy of Microbiology (EAM).
Metabolic engineering is a technology that designs microorganisms to produce desired substances in greater amounts and more efficiently by altering their genes. Put simply, it gives microorganisms new production capabilities, turning them into cell factories.
The problem is that living organisms are exceedingly complex. In the past, to check one by one which gene makes which protein and exactly what that protein does, researchers had to conduct experiments themselves.
Now AI predicts function from protein sequences. Lee began using AI in earnest for research in 2016. At the time, he said he assembled graphics processing units (GPUs) used in gaming computers himself to run Deep Learning models.
What he developed that way was the Deep Learning-based enzyme function prediction system DeepEC. Enzymes are proteins that drive various chemical reactions in living organisms. To turn microorganisms into factories that produce desired substances, you must first know precisely which enzymes play which roles.
The research team analyzed the functions of 34 million proteins using DeepEC. Doing the same task with the conventional standard method would take 150 years, but DeepEC finished it within 10 days.
After applying transformers, a cutting-edge AI technology, the accuracy improved further. Lee said, "AI learned on its own to predict even substrate-binding sites and cofactor-binding sites of proteins that it had not been taught," and added, "Through this, we confirmed the potential for AI to become a new tool for bio research."
The protein functions analyzed by AI led to actual substance production. Plastics are a prime example.
Lee said he began related research with the view that, because Korea has no crude oil, it must change its petro-dependent method of producing chemical materials.
The team redesigned the genes of Escherichia coli to make the cells stockpile plastic inside. The engineered microorganisms were able to produce about 180 grams of biodegradable plastic per liter.
Lee's team went beyond microbial research and also applied AI to predict adverse drug reactions.
Lee explained, "If you assume taking the 2,159 approved drugs in pairs, the combinations reach about 2.3 million, but the adverse reactions known so far are at the level of 460,000," adding, "It was impossible to know whether the remaining combinations had no side effects or simply had not been identified."
The team developed a system that analyzes drug structures using AI and predicts the possibility of 113 types of adverse reactions. It also proposed a method to identify alternative drugs that have similar effects but different structures, instead of drugs expected to pose side-effect risks.
Lee said, "In the case of the COVID-19 treatment Paxlovid, an analysis of interactions with 2,248 approved drugs showed potential drug interactions with 1,600."
The reason AI is needed in bio research is not only to speed up research. Lee believed the technology ultimately must be used to address humanity's larger problem of aging.
Lee said, "There are currently about 140 million dementia patients worldwide, and the number is expected to reach 400 million by 2050," adding, "As aging intensifies, we must develop new treatments, medicines, foods, and medical materials faster, but it is difficult to unravel all the complex systems of living organisms with human research capacity alone."
He said, "It has been confirmed that by using microorganisms, we can produce substances that were previously considered difficult to make," adding, "Going forward, we will not be able to think of bio, medicine, engineering, and AI as separate."