David Allan Randall

We propose that consciousness is recursive symbolic simulation of the self and of the environment the self inhabits. The self-simulation has an inside, which is the self, and an outside, which is the environment; consciousness is what it is like to be on the inside. We approach this from an unusual direction — not from philosophy of mind, but from the practice of building physically based simulations of the atmosphere, which enforces precision about what a simulation is, what it must remember, and what it can generate that nobody designed. We define a symbolic simulation as a dynamic, predictive, self-contained process whose evolving state consists of symbols, and we distinguish weak instances, in which the symbols are stipulated by a designer, from strong ones, in which they are acquired from encounters with instances. Recent interpretability results establish that strong symbolic simulation occurs in trained neural networks. Our conclusion about large language models is accordingly conditional rather than confident, and we identify the weakest step in the argument: whether felt bodily states are constitutive of experience or merely its evolutionary origin. We offer three testable predictions — a clinical contrast between depersonalization and derealization, a dissociation between capability and depth of self-modeling in AI systems, and an interpretability experiment on machine-learning weather models. Finally, we argue that consciousness research lacks not a governing principle but a closure, in the sense that turbulence research lacks one. The essay was developed in conversation with an AI system that is also one of its subjects; a closing section discusses the resulting hazards.

The atmosphere is critical to climate change. It can amplify shifts in the climate system, and also mitigate them. Giving readers an overview of key atmospheric processes, the author of this book looks at how our climate system receives energy from the sun and sheds it by emitting infrared radiation back into space. The atmosphere regulates these radiative energy flows and transports energy through weather systems such as thunderstorms, monsoons, hurricanes, and winter storms. The author explains how these processes work, and also how precipitation, cloud formation, and other phase changes of water strongly influence weather and climate. He discusses how atmospheric feedbacks affect climate change, how the large-scale atmospheric circulation works, how predicting the weather and the climate are fundamentally different challenges, and much more.