# Line Goes Up? Inherent Limitations of Benchmarks for Evaluating Large Language Models

arXiv: https://arxiv.org/abs/2502.14318 · HTML: https://mlsanity.com/paper/2502.14318/

## Key facts

- Cited by 14 papers in the map (data through 14 Oct 2026).
- First posted on arXiv 20 Feb 2025.

## Built on by (most cited first, 14 of 14)

- [Among Us: A Sandbox for Measuring and Detecting Agentic Deception](https://mlsanity.com/paper/2504.04072/) (2025)
- [RefactorCoderQA: Benchmarking LLMs for Multi-Domain Coding Question Solutions in Cloud and Edge Deployment](https://mlsanity.com/paper/2509.10436/) (2025)
- [Model-Native Computing Architecture: Envisioning Future System Architecture Through the Lens of Computer Architecture](https://mlsanity.com/paper/2606.00288/) (2026)
- [Rethinking the Understanding Ability across LLMs through Mutual Information](https://mlsanity.com/paper/2505.23790/) (2025)
- [Investigating Advanced Reasoning of Large Language Models via Black-Box Environment Interaction](https://mlsanity.com/paper/2508.19035/) (2025)
- [UniCode: Augmenting Evaluation for Code Reasoning](https://mlsanity.com/paper/2510.17868/) (2025)
- [Decomposing and Steering Functional Metacognition in Large Language Models](https://mlsanity.com/paper/2605.08942/) (2026)
- [Context Is Not Comprehension](https://mlsanity.com/paper/2506.04907/) (2025)
- [Testing chatbots on the creation of encoders for audio conditioned image generation](https://mlsanity.com/paper/2509.09717/) (2025)
- [SymSeqBench: a unified framework for the generation and analysis of rule-based symbolic sequences and datasets](https://mlsanity.com/paper/2512.24977/) (2025)
- [Theory Trace Card: Theory-Driven Socio-Cognitive Evaluation of LLMs](https://mlsanity.com/paper/2601.01878/) (2026)
- [Latent Performance Profiling of Large Language Models](https://mlsanity.com/paper/2605.30018/) (2026)
- [Using profiles of cognitive capability to assess AI suitability for workplace tasks](https://mlsanity.com/paper/2608.25623/) (2026)
- [Cost-Aware Best-LLM Identification using Dueling Feedback](https://mlsanity.com/paper/2609.30360/) (2026)

## Abstract (arXiv)

Large language models (LLMs) regularly demonstrate new and impressive performance on a wide range of language, knowledge, and reasoning benchmarks. Such rapid progress has led many commentators to argue that LLM general cognitive capabilities have likewise rapidly improved, with the implication that such models are becoming progressively more capable on various real-world tasks. Here I summarise theoretical and empirical considerations to challenge this narrative. I argue that inherent limitations with the benchmarking paradigm, along with specific limitations of existing benchmarks, render benchmark performance highly unsuitable as a metric for generalisable competence over cognitive tasks. I also contend that alternative methods for assessing LLM capabilities, including adversarial stimuli and interpretability techniques, have shown that LLMs do not have robust competence in many language and reasoning tasks, and often fail to learn representations which facilitate generalisable inferences. I conclude that benchmark performance should not be used as a reliable indicator of general LLM cognitive capabilities.