Circular Causality: A Short History (With Receipts) (medium.com)

The concept of circular causality, tracing its history from the mid-20th century, emphasizes feedback loops in understanding intelligent systems, including both minds and machines. Pioneers like Norbert Wiener, Warren McCulloch, and Gregory Bateson shifted the narrative from linear cause-effect relationships to recognizing the significance of circular feedback in biological and computational systems. For instance, Wiener defined cybernetics around circular causal systems, suggesting that intelligent behavior is driven by feedback loops, akin to a thermostat adjusting its heat based on prior performance. This redefinition allowed scientists to reconsider various fields through the lens of loops, fundamentally transforming how cognition and life are conceptualized. Today, this legacy persists in AI and Large Language Models (LLMs), which employ recursive loops during text generation but often lack persistent, self-sustaining states that mimic human cognition. While LLMs create temporary feedback cycles during interactions, they don't maintain ongoing internal circuits like a human mind or biological system. Innovations like AutoGPT attempt to incorporate recursive dynamics via external orchestration, pushing the boundaries of AI's capability to simulate circularity. Despite advancements, the AI community faces challenges in achieving true circular causality that reflects the depth of neurobiological processes, emphasizing the gap between current AI functionalities and genuine cognitive feedback systems.