Position: The Turing-Completeness of Autoregressive Transformers Relies Heavily on Context Management

Fuente: arXiv
Guardado en:
Detalles Bibliográficos
Autores principales: Cui, Guanyu, Wei, Zhewei, He, Kun
Formato: Preprint
Publicado: 2026
Materias:
Acceso en línea:
Etiquetas: Agregar Etiqueta
Sin Etiquetas, Sea el primero en etiquetar este registro!
_version_ 1866917539947216896
author Cui, Guanyu
Wei, Zhewei
He, Kun
author_facet Cui, Guanyu
Wei, Zhewei
He, Kun
contents Many works make the eye-catching claim that Transformers are Turing-complete. However, the literature often conflates two distinct settings: (i) a fixed Transformer system setting, in which a fixed autoregressive Transformer is coupled with a fixed context-management method to process inputs of different lengths step by step, and (ii) a scaling-family setting, in which a family of different models (with increasing context-window length or numerical precision) is used to handle different input lengths. Existing proofs of Transformer Turing-completeness are frequently established in setting (ii), whereas real-world LLM deployment and the standard notion of Turing-completeness correspond more naturally to setting (i). In this paper, we first formalize the fixed-system setting, thereby providing a concrete characterization of how real-world LLMs operate. We then argue that results proved in the scaling-family setting provide theoretically meaningful resource bounds but do not establish Turing-completeness, thereby clarifying a common misinterpretation of existing results. Finally, we show that different context-management methods can yield sharply different computational power, and we advocate the position that context management is a central component that critically determines the computational power of real-world autoregressive Transformers.
format Preprint
id arxiv_https___arxiv_org_abs_2605_19514
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle Position: The Turing-Completeness of Autoregressive Transformers Relies Heavily on Context Management
Cui, Guanyu
Wei, Zhewei
He, Kun
Artificial Intelligence
Computation and Language
Machine Learning
Many works make the eye-catching claim that Transformers are Turing-complete. However, the literature often conflates two distinct settings: (i) a fixed Transformer system setting, in which a fixed autoregressive Transformer is coupled with a fixed context-management method to process inputs of different lengths step by step, and (ii) a scaling-family setting, in which a family of different models (with increasing context-window length or numerical precision) is used to handle different input lengths. Existing proofs of Transformer Turing-completeness are frequently established in setting (ii), whereas real-world LLM deployment and the standard notion of Turing-completeness correspond more naturally to setting (i). In this paper, we first formalize the fixed-system setting, thereby providing a concrete characterization of how real-world LLMs operate. We then argue that results proved in the scaling-family setting provide theoretically meaningful resource bounds but do not establish Turing-completeness, thereby clarifying a common misinterpretation of existing results. Finally, we show that different context-management methods can yield sharply different computational power, and we advocate the position that context management is a central component that critically determines the computational power of real-world autoregressive Transformers.
title Position: The Turing-Completeness of Autoregressive Transformers Relies Heavily on Context Management
topic Artificial Intelligence
Computation and Language
Machine Learning
url https://arxiv.org/abs/2605.19514