Multispin Physics of AI Tipping Points and Hallucinations

TL;DR

This paper models AI hallucination tipping points using multispin physics, revealing how prompt choices and architecture amplify risks, with implications for transparency, safety, and liability.

Contribution

It introduces a novel multispin thermal system model of AI tipping points, providing an exact formula to predict and analyze hallucination risks.

Findings

Identifies a hidden tipping instability at the AI's attention head level.

Derives an exact formula for the AI tipping point influenced by prompts and biases.

Shows how multilayer architecture amplifies hallucination risks.

Abstract

Output from generative AI such as ChatGPT, can be repetitive and biased. But more worrying is that this output can mysteriously tip mid-response from good (correct) to bad (misleading or wrong) without the user noticing. In 2024 alone, this reportedly caused $67 billion in losses and several deaths. Establishing a mathematical mapping to a multispin thermal system, we reveal a hidden tipping instability at the scale of the AI's 'atom' (basic Attention head). We derive a simple but essentially exact formula for this tipping point which shows directly the impact of a user's prompt choice and the AI's training bias. We then show how the output tipping can get amplified by the AI's multilayer architecture. As well as helping improve AI transparency, explainability and performance, our results open a path to quantifying users' AI risk and legal liabilities.