Automated Scientific Discovery: Papers from the AAAI Fall Symposium
Selmer Bringsjord and Andrew Shilliday, Cochairs
There is a long and fascinating history of humankind's endeavor to explain and, with the advent of AI, ultimately mechanize the overarching processes that lead to scientific discoveries. This quest dates back to Aristotle's account of human deductive reasoning (the theory of the syllogism, developed to model the discoveries of Euclid), and continues through modern AI, which, through impressive systems like LT, Bacon, GT, Eurisko, and Graffiti (and many theorem provers, model finders, and computational frameworks for machine-assisted reasoning), has placed some degree of such automation within reach. Over the past 60 years, starting with AI's inaugural conference, systems such as these have automated aspects of scientific discovery. Machines have generated novel and interesting conjectures (some of which have spawned new scientific research areas), and increasingly efficient techniques have been invented to prove or refute them. Nevertheless, the sobering fact remains that such advances fall far short of approaching the creativity and innovation of even amateur scientists. We believe that AI is ripe for revolutionary progress in automated and semiautomated scientific discovery, in no small part because the field now has on hand systems that mark advances in various parts of discovery — parts that, when interconnected, may make for some very exciting new systems. We also believe that dialogue between researchers behind these systems could lead to a new generation of powerful AI discovery systems.