Abductive Reasoning, Which AI Cannot Perform, Is Not Taught in Postgraduate Programs
A Google DeepMind report shows that artificial intelligence is proficient in induction and deduction, but lacks abduction. Experts state that business training programs must be redesigned to close this gap.
An evaluation report published by Google DeepMind revealed that while artificial intelligence succeeds in induction and deduction processes, it lacks abduction—the capacity to generate novel explanations based on surprising observations. Experts emphasize that postgraduate programs like MBAs focus on tasks that AI can easily perform, but fail to teach this critical cognitive leap that the human mind should make.
A January article written by Tom Zahavy of Google DeepMind questioned whether today's AI could have invented general relativity if it had been given all of Albert Einstein's knowledge from 1907, and concluded that it would not. According to the article, machines are rapidly improving at finding patterns in data (induction) and drawing conclusions from premises (deduction). However, what they lack is abduction—the ability to generate a new explanation for a surprising observation, which is the skill of making an uncompressed, sudden leap from data, much like Albert Einstein's thought experiment about a free-falling observer.
Dr. Phuong Nguyen, an executive education instructor at CFVG, states that executive development programs reward students for recognizing case frameworks and deriving conclusions, yet these processes are areas where machines dominate. According to Nguyen, the actual deficiency is the failure to provide education in abduction, which is the phase of realizing that a situation does not fit the mold and generating an explanation that no one else has offered.
It is recalled that this cognitive leap was methodologically formulated years ago through grounded theory work by sociologists Barney Glaser and Anselm Strauss in 1967. According to the work of sociologists such as Stefan Timmermans and Iddo Tavory, the engine of this theory is not induction, but rather an abductive process conducted with a prepared mind.
It is stated that the fundamental question educational programs must ask when integrating artificial intelligence is which cognitive actions should be reserved for students. It is argued that students should write down their own positions before consulting any tool, make predictions prior to making a decision, and record evaluations in their own words.