When discussing the impact of AI on radiology, just remember: radiology is an infinite game

Abstract

Over the past few years, headlines suggesting the demise of radiology under the rise of artificial intelligence (AI) have spread in specialized and lay media. One of the most emblematic testimonials came from Professor Geoffrey Hinton of the University of Toronto, an iconic researcher on neural networks and machine learning. In 2016, professor Hinton stated that "people should stop training radiologists now" and "it is just completely obvious that within 5 years deep learning will do better than radiologists" (1). The impact of the bad news has been noticed by medical students. A considerable proportion of medical students have been discouraged from considering radiology solely because of the uncertain impact of AI on the field (2,3). Indeed, the notion that AI may negatively impact radiology in the near future permeates the medical imaging community (4). When such discussion emerges, many of us forget one of the most important factors when analyzing the potential demise of radiology in the face of AI: radiology is an infinite game. The concept of finite and infinite games was introduced in the 1980s by author James P. Carse (5). A finite game, among other characteristics, has known players, known rules (rules to which players have agreed upon) and a clear definition of when the game ends. Players play to win. Many sports, including football and hockey, are finite games. There is a beginning, a middle, and an end, with a winner and a loser. In contrast, infinite games are those with known and unknown players, no exact set of rules, in which a player leaves the game only when low on resources or no longer willing to play. Players play to keep playing (i.e., to perpetuate the game). There is no finish line, no winner, and no loser. This concept has been extensively analyzed by Simon Sinek in the business world (6) , with many examples of the application of this concept in the real world. It is no different in radiology. We most often do not know all the players (radiology groups, radiology practices, non-radiology physicians, hospitals, etc.). The players determine their own strategies, and, other than ethical, moral, and legal guidelines, which may vary from country to country, there are no fixed rules to which everyone