- There is no foolish study... Korean scientists should be more bold-
"When I developed 'Chematica', a lot of criticism said it is 'unnecessary research'. If I was afraid of failure, I would never have completed it,"
Said Bartosz Grzybowski, a Group Leader (distinguished professor of Ulsan National Institute of Science and Technology, photo) in the Center for Soft and Living Matter, within the Institute for Basic Science (IBS), at the Strong Korea Forum 2018 held at the Millennium Hilton Hotel in Seoul on March 31.
Professor Grzybowski is a chemist who astounded academics by developing an artificial intelligence (AI) for chemical synthesis called 'Chematica' in 2012, which is called 'Alpha Go' of the chemical industry. In recognition of this achievement, in 2016, he received the "Feynman Prize", a prestigious award in nanotechnology.
Chematica is an AI program that synthesizes chemicals and guides optimal routes. It has learned all 30,000 chemical synthesis methods accumulated since 1700s by human, and calculate the optimal synthesis method at a much faster speed than humans. Merck, a German pharmaceutical company, took over GSI (솓 developer) last year, looking at the potential of this program. GSI is a company founded in the United States named by Prof. Grzybowski himself.
"Applying AI to chemistry is much harder than making AI play chess or go," he said. Unlike chess or go, which only consider its future move, chemical synthesis must take into account various variables such as molecular bonding state and three-dimensional molecular structure.
"Considering variables, 100th power of 100 cases occur even when synthesize simple compounds," he said. "The complexity is why many people thought Chematica would fail."
Professor Grzybowski predicted that AI would revolutionize pharmaceutical field. It is expected that the synthetic method developed by AI instead of humans can produce drugs more efficiently. AI can also be used to circumvent the patents of big pharmaceutical companies.
"By using AI, we were able to find a synthetic method that was more efficient than the existing method and avoided patent problems in just 10 minutes," said Prof. Grzybowski. "The concept of intellectual property (IP) would be completely changed."
Asked whether AI could replace scientists, he decisively said, "Never." He emphasized that only people can pioneer new fields.
He also gave some advice to the Korean scientific community. "Korean students tend to memorize knowledge," he said. "In the AI era, conceptualization is more important than memorization. It is fortunate that Korea has been investing heavily in basic science for last several years," he said. "I hope that Korean scientists will be actively challenging new research with this support."