Contemplative Wisdom for Superalignment

As artificial intelligence (AI) improves, traditional alignment strategies may falter in the face of unpredictable self-improvement, hidden subgoals, and the sheer complexity of intelligent systems. Rather than externally constraining behavior, we advocate designing AI with intrinsic morality built into its cognitive architecture and world model. Inspired by contemplative wisdom traditions, we show how four axiomatic principles can instil a resilient Wise World Model in AI systems. First, mindfulness enables self-monitoring and recalibration of emergent subgoals. Second, emptiness forestalls dogmatic goal fixation and relaxes rigid priors. Third, non-duality dissolves adversarial self-other boundaries. Fourth, boundless care motivates the universal reduction of suffering. We find that prompting AI to reflect on these principles improves performance on the AILuminate Benchmark using GPT-4o, particularly when combined. We offer detailed implementation strategies for state-of-the-art models, including contemplative architectures, constitutions, and reinforcement of chain-of-thought. For future systems, the active inference framework may offer the self-organizing and dynamic coupling capabilities needed to enact these insights in embodied agents. This interdisciplinary approach offers a self-correcting and resilient alternative to prevailing brittle control schemes.

Causal blankets: Theory and algorithmic framework

We introduce a novel framework to identify perception-action loops (PALOs) directly from data based on the principles of computational mechanics. Our approach is based on the notion of causal blanket, which captures sensory and active variables as dynamical sufficient statistics -- i.e. as the "differences that make a difference." Moreover, our theory provides a broadly applicable procedure to construct PALOs that requires neither a steady-state nor Markovian dynamics. Using our theory, we show that every bipartite stochastic process has a causal blanket, but the extent to which this leads to an effective PALO formulation varies depending on the integrated information of the bipartition.