Language processing in humans and computers
This is an incremental analysis of hallucination issues in language models, relevant for AI safety and cognitive science.
The paper examines how language models, after learning to recognize hallucinations and dream safely, generate broader systems of false beliefs and self-confirming theories, similar to human tendencies.
Machine-learned language models have transformed everyday life: they steer us when we study, drive, manage money. They have the potential to transform our civilization. But they hallucinate. Their realities are virtual. This note provides a high-level overview of language models and outlines a low-level model of learning machines. It turns out that, after they become capable of recognizing hallucinations and dreaming safely, as humans tend to be, the language-learning machines proceed to generate broader systems of false beliefs and self-confirming theories, as humans tend to do.