It is argued that AI developers should track and report the NLS depth of their models as a complement to direct measurements of CoT monitorability, and it is found that, across open-weight models, NLS depth correlates strongly with layer count, with some exceptions, and has grown slowly over recent years.
Abstract
Chain-of-thought (CoT) is currently a valuable tool for overseeing AI models, but some architectural shifts could significantly reduce CoT monitorability. We operationalize a measure that serves as a proxy for the amount of unverbalized serial cognition a model can perform, as a specific instantiation of the notion of opaque serial depth: roughly, the length of the longest path through a model's computation that does not pass through an interpretable bottleneck. Our main contribution is a particular standard for what counts as such a bottleneck. We define natural-language-rooted (NL-rooted) nodes as nodes that were initialized from a pre-training prior to produce human-understandable outputs, whose output space has not been expanded beyond that prior, and whose outputs the model has not been modified to interpret as a different type of data. By construction, the resulting measure, NLS depth, can be adjudicated before training from an architecture, training recipe, and deployment configuration alone, and is insensitive to ordinary post-training such as prompting, distillation, and RL. We find that, across open-weight models, NLS depth correlates strongly with layer count, with some exceptions, and has grown slowly over recent years. It would likely grow slowly under continued scaling of standard transformers, while latent-reasoning architectures such as full-bandwidth transformers could increase it by a large amount. We work through how NLS depth applies to a range of existing and proposed architectures, discuss alternative operationalizations, and give a worked example bounding the depth of Gemma3-1B. We argue that AI developers should track and report the NLS depth of their models as a complement to direct measurements of CoT monitorability.
The ceiling of a task therefore answers when a transformer can succeed without its chain of thought, and the best accuracy that a shallow circuit can reach on a task the ceiling of the task, and a task is serial if its ceiling lies below one.
Large language models need several orders of magnitude more text than a child hears to reach a comparable level of linguistic skill. One reason is that a particularly useful capability, the induction head (a circuit that completes a pattern of the form [A] [B] ... [A] by copying [B]), is not built in but must emerge du...
Josh Taylor· Proceedings of the Second Ba...· 0 citations
Large language model (LLM) agents achieve state-of-the-art performance across domains through chain-of-thought (CoT) reasoning. Despite their ubiquitous use, little is understood about the relationship between the externalized trace and the model's actual computation. We address this by investigating how LLM agents mak...
Chaipat Tirapongprasert, Rajarshi Das, P. Dube· 0 citations
Agents based on large language models no longer answer a prompt. They plan, call tools, and keep memory across steps. That creates engineering problems that a plain text generator never faces. For example, how much context should be kept? What happens if a provider fails halfway through a task? How can a failed step be...
R. E, Mokshita N, Ambika· Advanced International Journ...· 0 citations
SEER (Self-Enhancing Efficient Reasoning), a self-enhancing framework for adaptive CoT compression, improves the conciseness of reasoning while preserving output quality, without relying on external compression tools.
Ke-Rui Huang, Shu-Han Liu, Xing Hu et al.· Proceedings of the ACM on So...· 9 citations· ⚡2
Related blog posts
MIT News · Artificial Intelligence· news.mit.eduOct 13, 2026
Machine-learning networks can learn without detailed instructions from a designer. Those that do may offer insight into how the brain reorganizes itself during learning.
MIT News · Artificial Intelligence· news.mit.eduSep 29, 2026
Professor Sherry Turkle’s new book, “Artificial Intimacy,” offers a withering critique of chatbots and the antisocial dynamics she believes they encourage.