Creating the Engine for Scientific Discovery as an Ultimate Grand Challenge: The Purpose of SBI/SBX Group
Scientific discovery has been a driving force of our civilization. Accelerating scientific discovery is one of the most important missions that help shape the future of our society. Systems Biomedicine is one of the areas that shall benefit largely from acceleration of scientific discovery. With the complexity of biology and vastness of data, systems biology and systems biomedicine research has been facing serious bottlenecks in recent years. Development of platforms that automate research processes and even developing AI Scientist at high-level of autonomy shall be a major breakthrough.
More than 20 years has been passed since the emergence of systems biology with expectation of understanding biological systems at the system-level through developing large-scale and detailed computational models. Unfortunately, this ambition is yet to be realized partly due to cognitive and sociological limitations of research driven by human researchers. Cognitive limitations prevent us from properly handling large-scale complex and dynamical systems with massive data. Issues arising cognitive limitations are described in an article published in AI Magazine 2016 (Kitano, H., “Artificial Intelligence to Win the Nobel Prize and Beyond: Creating the Engine for Scientific Discovery”, AI Magazine, 2016: URL: https://ojs.aaai.org//index.php/aimagazine/article/view/2642). Through over two decades of systems biology research, I am convinced that developing “AI Scientist” is the best way to transform systems biology into the next stage to achieve its objective.
At the same time, automating scientific discovery requires in-depth understanding of the process of scientific discovery, either it is achieved by human or machine. Our working hypothesis here is that scientific discovery can be defined as a process of massive hypothesis generation and verification. Human scientists try to generate most plausible hypothesis and verify them experimentally at a shortest path. The question is how we know a priori which hypothesis is more plausible and have higher value than other hypotheses. The radical solution shall be generating hypotheses exhaustively, computationally filtering them with simulation and logical consistency, and verify them using massive automated experimental facilities. Readers may wish to take a look at my recent article for more details on this discussion. (Kitano, H., “Nobel Turing Challenge: creating the engine for scientific discovery”, npj Systems Biology and Applications, 2021: URL: https://www.nature.com/articles/s41540-021-00189-3)
The grand challenge is to develop AI Scientist that can make major scientific discoveries highly autonomously some of which may worth Nobel Prize or beyond. The Nobel Turing Challenge is proposed and a workshop to promote the challenge was already hosted by the Alan Turing Institute, London late Feb. 2020 organized by Prof. Ross King, Prof Yolanda Gil, and myself. (The workshop report “AI Scientist Grand Challenge: Summary of discussion during workshop held in February 2020” can be found here: https://www.turing.ac.uk/sites/default/files/2021-02/summary_of_discussion_workshop_2020_ai_scientist_grand_challenge_clean.pdf) The workshop was organized as a part of a research project at the Alan Turing Institute fro the Nobel Turing Challenge (https://www.turing.ac.uk/research/research-projects/turing-ai-scientist-grand-challenge).
Pushing the boundaries of scientific discoveries in life sciences is our goal, and we are determined to achieve this by providing a platform for automation of research and autonomous systems for biomedical discoveries. Therefore, we defined our mission: “transforming our civilization by developing the engine of scientific discovery.” Accomplishing this mission requires collective efforts of multiple and broad range of partners. As a first step, we have built platforms for connecting resources and kick starting the discovery process that are; Garuda Connectivity and Automation Platform, Taxila Text Mining and Knowledge Extraction Platform, and Gandhara AI framework. These platforms from a basic infrastructure where various purpose specific systems can be developed. Our platforms and custom solutions provide highly intelligent services to assist scientific activities in drug discovery, clinical trials, healthcare and in domains beyond biology.
Artificial Intelligence to Win the Nobel Prize and Beyond: Creating the Engine for Scientific…
Hiroaki Kitano Sony Computer Science Laboratories https://doi.org/10.1609/aimag.v37i1.2642 Abstract This article…