Abstract
Artificial Intelligence (AI), even prior to its inception many decades ago has aroused both fear and excitement. The misconception that “intelligent” artifacts should necessarily be human-like has largely blinded society to the fact that for many decades industrial machines have been achieving artificial intelligence and automate repetitive tasks. AI works by combining large amounts of data with fast, iterative processing and intelligent algorithms, allowing the software to learn automatically from patterns or features in the data. In the last decade there were many applications of AI in a wide range of activities including medical diagnosis, electronic trading platforms, robotics, remote sensing and playing games like chess. AI has been used in numerous fields and industries, including finance, healthcare, education, and transportation. A famous example was IBM’s Deep Blue supercomputer defeating famous chess master player Gary Kasparov in May 1997. In 2017 AlphaZero (created by Google sibling DeepMind, an AI game-playing programme) within 24 hours of training achieved a superhuman level of play chess by defeating world-champion program Stockfish (one of the strongest chess engines than the best human chess grandmasters.). But AI has successful applications in science and in particular for new drug discovery. In 2019, researchers in the Massachusetts Institute of Technology (MIT) used a pioneering machine-learning approach (machine-learning algorithm) to identify new types of antibiotic from a pool of more than 100 million molecules. Out of this process the AI programme found one, Halicin, that works against a wide range of bacteria, including tuberculosis and strains considered untreatable. It is the first antibiotic discovered with artificial intelligence (AI). Although AI has been used to aid parts of the antibiotic-discovery process before, this was the first time it has identified completely new kinds of antibiotic from scratch, without using any previous human assumptions. The idea of using predictive computer models for “in silico” screening is not new, but until now, these models were not sufficiently accurate to transform drug discovery. The new neural networks can learn these representations automatically, mapping molecules into continuous vectors which are subsequently used to predict their properties.