No. 114/2022

The Chinese professor of computer science, Yaochu Jin, is trying to apply the evolutionary principle to AI and thus expand its potential uses into fields like vaccine research. To do so, he trains the AI to organise itself and to deal with tasks that are not clearly defined – that is, to respond flexibly to new information and optimise itself over time. Together with colleagues in veterinary medicine, Jin has developed an evolutionary learning algorithmwhich is supposed to help fight the foot and mouth disease that afflicts livestock. Just like the Corona virus, these pathogens constantly mutate. “In order to find out whether a vaccine is still effective, we normally have to do tests on animals, which are expensive and ethically questionable,” says Jin. Instead, the AI can compare the amino acids in the proteins of the new and old variants of a virus. “It recognises howmany of the perhaps 300 decisive positions in the protein have changed and how relevant that is. It then predicts the effectiveness of the vaccine.” Research has already achieved something similar using AI on flu and Ebola vaccines, says Jin. It could prove a promising approach to Corona, too. Text JAN BERNDORFF HOW CAN ARTIFICIAL INTELLIGENCE HELP TO COMBAT VIRUSES, MR JIN? Photo: Humboldt Foundation/Elbmotion BRIEF ENQUIRIES Being faster and more reliable than humans, artificial intelligence is usually used to eval- uate huge volumes of data and identify pat- terns in them. But what happens when there isn’t enough data, and you have to work with uncertainties? In this case, thanks tomillions of years of evolution, nature easily has the upper hand. KEYWORDS: medicine, swarm robotics In autumn 2021, the Chinese computer scientist, YAOCHU JIN, relocated from the British University of Surrey to Bielefeld University as an Alexander von Humboldt Professor for Artificial Intelligence. 8 HUMBOLDT KOSMOS 114/2022