Dualism

Dualism in AI highlights the fundamental distinction between two primary approaches to artificial intelligence: symbolic AI and sub-symbolic AI. Symbolic AI, often associated with early AI research, relies on high-level human-readable symbols and rules to represent knowledge and perform reasoning. In contrast, sub-symbolic AI, such as neural networks and other machine learning methods, operates at a lower level, learning patterns from data without explicit symbolic representation. This dualism represents the differing philosophies and methodologies in AI development, with symbolic AI focusing on logic and formal reasoning, while sub-symbolic AI emphasizes learning from experience and data-driven approaches. The integration and interaction between these two paradigms are crucial for creating more versatile and effective AI systems.

The concept of dualism in AI became prominent in the mid-1980s when the rise of connectionist models, particularly neural networks, began to challenge the dominance of symbolic AI. Symbolic AI had been the mainstay since the inception of AI in the 1950s, but the resurgence of sub-symbolic methods in the 1980s marked a significant shift.

Notable figures in the development of symbolic AI include Allen Newell and Herbert A. Simon, who worked on symbolic reasoning and cognitive architectures. On the sub-symbolic side, Geoffrey Hinton, David Rumelhart, and Ronald J. Williams made significant contributions with their work on neural networks and backpropagation. The interplay between these approaches has been shaped by many researchers across decades, reflecting the evolution and maturation of the AI field.