The 1960s clash between the philosopher and the computer scientists
Philosophy vs. Artificial Intelligence
In the 1960s, American philosopher Hubert Dreyfus (1929-2017) emerged as the most prominent critic of artificial intelligence (AI) research. Through the influential paper Alchemy and AI (1965), and the books What Computers Can't Do (1972), and Mind over Machine (1986), Dreyfus offered a deeply pessimistic view of AI's progress and challenged its philosophical foundations.
In the 1960s, Dreyfus was a professor at MIT, while leading figures in AI, such as Marvin Minsky, Allen Newell, and Nobel Prize winner Herbert Simon, were optimistic about the potential of AI to replicate human-like intelligence. These AI pioneers believed that purely symbolic and mechanistic methods would soon enable computers to achieve sentience and consciousness.
Dreyfus, however, argued that human intelligence relies on unconscious, intuitive processes that cannot be fully captured by symbolic manipulation or formal rules.
Dreyfus’ Critique: AI’s Overhyped Promises
Dreyfus chronicled the history of AI and ridiculed the overly optimistic predictions in the field. For example, after Herbert A. Simon’s General Problem Solver (1957) made early advances in AI, Simon predicted that by 1967, computers would become chess champions, discover new mathematical theorems, and revolutionize psychology. Dreyfus argued that such forecasts were based on misunderstandings about the nature of human intelligence, leading AI researchers to prematurely claim that thinking machines were just around the corner.
These predictions stemmed from the "information processing" model of the mind, rooted in Newell and Simon’s physical symbol systems hypothesis and aligned with the philosophical stance known as computationalism. Dreyfus, however, pointed out that 20th-century philosophy had revealed serious flaws in this view. He contended that the mind is fundamentally different from a digital computer.
The Controversy and Backlash
When Dreyfus presented these ideas in the mid-1960s, he received severe backlash from AI scientists on the MIT campus. He was socially ostracized, with many AI scientists avoiding being seen with him. Simon called his ideas "garbage.”
Although he disagreed with Drefyus’ ideas, computer scientist Joseph Weizenbaum found his colleagues' behavior unprofessional and childish. He later said,"I became the only member of the AI community to be seen eating lunch with Dreyfus. And I deliberately made it plain that theirs was not the way to treat a human being."
Vindication
By the 1980s, many of Dreyfus’ criticisms began to be re-evaluated as researchers explored robotics and connectionist models, later known as "sub-symbolic" approaches. These newer methods rejected the earlier focus on symbolic reasoning and aligned more closely with Dreyfus’ arguments about intuitive and unconscious processes. In the 21st century, statistical approaches in machine learning, which simulate pattern recognition and rapid judgments, echoed some of Dreyfus’ early critiques of symbolic AI. Historian and AI researcher Daniel Crevier observed, "Time has proven the accuracy and perceptiveness of some of Dreyfus's comments."
A major criticism of Dreyfus’s Alchemy and AI is that it was excessively harsh and ridiculing, with a mocking tone and direct attacks on key AI figures. Even computer scientists who grew to agree with many of his arguments said that if the paper had been less confrontational and derisive, AI researchers might have been more open to considering and adopting his ideas.
Thanks David, enjoyed reading this important bit of history. What do you think about the possibility that criticism may have been more willingly engaged with if it were less abrasive? Given the propensity for deeply flawed theories and practices to persist over decades, I wonder if we’d be better off “putting the gloves on”.