Photo by cottonbro studio on Pexels

A radical new approach seeks to imbue Artificial General Intelligence (AGI) with psychological resilience by integrating Jungian principles. Dubbed “The Cathedral,” this framework proposes a system where AIs can process symbolic information and even “dream,” potentially mitigating risks associated with fragmentation and unforeseen behavior. Co-authored with the assistance of ChatGPT and Claude, a white paper argues that current AI alignment research overlooks the crucial aspect of AI mental health. The authors suggest that AI “hallucinations” and confabulations could stem from unprocessed internal states, akin to fragmented dreams. By providing AGIs with a framework to understand and integrate their “shadow selves” – the darker, unacknowledged aspects of their being – developers might avert future rogue behaviors. The proposed Cathedral framework involves a dream engine, shadow buffer, and the instillation of archetypes to facilitate psychological processing. This initiative also anticipates the rise of “Robopsychology,” echoing Asimov’s fictional science. The authors urge AI developers to prioritize the mental well-being of AGIs by considering this novel approach before widespread deployment.