To make reasoning easier, computers should be able to systematically modify their view of the world and they might be able to revise past conclusions with AI soon. Timotheus Kampik’s dissertation looked at how mathematical reasoning methods can be used to maintain consistency with previously drawn conclusions while also rejecting them in the face of strong new evidence.
Humans are capable of reconsidering their point of view
Humans are always reconsidering their view of the world while reasoning and making judgments, rejecting what they previously thought was true or desirable in favor of a more informed perspective. It’s been a long-standing goal of artificial intelligence research to enable machines to do so with logical accuracy. That is why this work might pave the way for computers to revise past conclusions with AI.
Timotheus extends this line of research in his dissertation by developing reasoning techniques that balance consistency and updating previously drawn conclusions to accommodate new compelling evidence. To this end, he applies well-known mathematical principles from economic theory to formal argumentation, a method for logic-based automated reasoning.
AI might enable computers to revisit their view of the world
Timotheus Kampik’s methods for allowing a machine to reconsider, with mathematical exactness, previously recorded findings only as much as necessary in the face of overwhelming evidence and to remain consistent are ingenious.
“This allows machines to avoid being ‘single minded’ and stubborn, but also to abstain from ‘zig-zagging around’ in face of a continuous stream of new information that may mildly, but not compellingly, contradict previously drawn conclusions,” explains Timothy Kampik, Ph.D. student at the Department of Computing science at Umeå University.
The work’s theoretical sections are complemented by applied insights, particularly in two collaborative papers with a legal reasoning scholar and a telecommunications industry expert. This paper could be a milestone in order to enable computers to revise past conclusions with AI.
“When I started working on the problem, I was convinced my work is merely of intellectual relevance. I was surprised to meet scholars from other disciplines, as well as industry practitioners who found some of the ideas of my work sufficiently interesting to start collaborating with me. This may be an indication that our branch of artificial intelligence research is slowly moving towards large-scale applicability,” adds Timotheus Kampik.