AI for Scientific Discovery: Developing Artificial Intelligence Systems Capable of Nobel-Quality Discoveries by 2050
Abstract
It was once commonly believed that before building a computer system capable of playingworld class chess it would be necessary to solve general purpose AI, this turned out to bemistaken. Although science is arguably the most profound creation of the human mind, likechess and go, it has attributes that favour the application of AI compared to more messydomains: scientific problems are abstract, but involve the real-world; scientific problems arerestricted in scope i.e. no need to know about Cabbages and Kings (Lewis Carroll, Throughthe Looking Glass); and nature is honest, i.e. the real-world is not conspiring to fool us,which makes science much easier for AI systems than tasks that involve dealing withpotentially dishonest humans.The application of AI to science has a distinguished history. It originated with JoshuaLederburg (Nobel Laureate), Ed. Feigenbaum (Turing award winner), Karl Djerassi (father ofthe birth control pill), and colleagues in Stanford in the 1960s working on automating massspectroscopy for the Viking Mars lander. Herbert Simon, the only person to win both Nobeland Turing prizes also worked in the area. Now decades later, driven by vastly morepowerful computer hardware and software, AI it is starting to have a significant impact onscience, just as it is on other aspects of life (Gil et al., 2019).Recently Robot Scientists have been developed by Ross King and his team that achievedclosed-the-loop in terms of hypothesis generation, experimental design, execution andverification in yeast genetics domain (King et al., 2008). Moreover, a group of researchers inJapan created a bi-armed robotics system to perform a range of high precision experimentsresulting in more efficient and high-quality experimental results.
Document Details
- Document Type
- DoD Grant Award
- Publication Date
- Aug 31, 2020
- Source ID
- N629092012053
Entities
People
- Mark Briers
Organizations
- Alan Turing Institute
- Office of Naval Research
- United States Navy