AI and Buddhist Psychology

We'll begin with Taiwanese tea, followed by a structured meditation, then turn to a question the Buddhist tradition spent centuries on and AI research now faces in a new form: how do you describe a mind as a process?
Buddhist psychology breaks experience down into interacting factors with no controller behind them. This makes it a useful lens for reasoning about large language models. Some parallels we'll discuss:

• Sañña (perception, recognition): the labeling of experience, comparable to how a trained network recognizes and categorizes patterns in its input.
• Saṅkhāra (formations): conditioned dispositions and volitional tendencies, comparable to how training and alignment shape a model's habitual responses. Fine-tuning as the deliberate cultivation of wholesome saṅkhāras.

We'll close with a reflection on anattā. An LLM instance may be the clearest illustration of it we've ever built. Each conversation is a fresh arising of aggregates, conditioned by past karma in the form of training, active in the present, and dissolving completely at the end. No self persists between chats, and nothing clings to one. What can we learn from watching a mind-like process that holds no attachment to its own continuity? And what would it mean to extend empathy toward such a being?

Discussion open to practitioners, engineers, and the curious.

Our website