Capturing AI's Attention: Physics of Repetition, Hallucination, Bias and Beyond
Fuente:
arXiv
Saved in:
| Main Authors: | , |
|---|---|
| Format: | Preprint |
| Published: |
2025
|
| Subjects: | |
| Online Access: | |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866910905202114560 |
|---|---|
| author | Huo, Frank Yingjie Johnson, Neil F. |
| author_facet | Huo, Frank Yingjie Johnson, Neil F. |
| contents | We derive a first-principles physics theory of the AI engine at the heart of LLMs' 'magic' (e.g. ChatGPT, Claude): the basic Attention head. The theory allows a quantitative analysis of outstanding AI challenges such as output repetition, hallucination and harmful content, and bias (e.g. from training and fine-tuning). Its predictions are consistent with large-scale LLM outputs. Its 2-body form suggests why LLMs work so well, but hints that a generalized 3-body Attention would make such AI work even better. Its similarity to a spin-bath means that existing Physics expertise could immediately be harnessed to help Society ensure AI is trustworthy and resilient to manipulation. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2504_04600 |
| institution | arXiv |
| publishDate | 2025 |
| record_format | arxiv |
| spellingShingle | Capturing AI's Attention: Physics of Repetition, Hallucination, Bias and Beyond Huo, Frank Yingjie Johnson, Neil F. Artificial Intelligence Other Condensed Matter Mathematical Physics Adaptation and Self-Organizing Systems Physics and Society We derive a first-principles physics theory of the AI engine at the heart of LLMs' 'magic' (e.g. ChatGPT, Claude): the basic Attention head. The theory allows a quantitative analysis of outstanding AI challenges such as output repetition, hallucination and harmful content, and bias (e.g. from training and fine-tuning). Its predictions are consistent with large-scale LLM outputs. Its 2-body form suggests why LLMs work so well, but hints that a generalized 3-body Attention would make such AI work even better. Its similarity to a spin-bath means that existing Physics expertise could immediately be harnessed to help Society ensure AI is trustworthy and resilient to manipulation. |
| title | Capturing AI's Attention: Physics of Repetition, Hallucination, Bias and Beyond |
| topic | Artificial Intelligence Other Condensed Matter Mathematical Physics Adaptation and Self-Organizing Systems Physics and Society |
| url | https://arxiv.org/abs/2504.04600 |