# Testing AI on language comprehension tasks reveals insensitivity to underlying meaning

arXiv: https://arxiv.org/abs/2302.12313 · HTML: https://mlsanity.com/paper/2302.12313/

## Key facts

- Cited by 45 papers in the map (data through 14 Oct 2026).
- Builds on 24 of its 79 references that are in the map.
- First posted on arXiv 23 Feb 2023.

## Builds on (most cited first, 24 of 24)

- [Language Models are Few-Shot Learners](https://mlsanity.com/paper/2005.14165/) (2020)
- [Training language models to follow instructions with human feedback](https://mlsanity.com/paper/2203.02155/) (2022)
- [Measuring Massive Multitask Language Understanding](https://mlsanity.com/paper/2009.03300/) (2020)
- [PIQA: Reasoning about Physical Commonsense in Natural Language](https://mlsanity.com/paper/1911.11641/) (2019)
- [A Broad-Coverage Challenge Corpus for Sentence Understanding through Inference](https://mlsanity.com/paper/1704.05426/) (2017)
- [Beyond the Imitation Game: Quantifying and extrapolating the capabilities of language models](https://mlsanity.com/paper/2206.04615/) (2022)
- [A large annotated corpus for learning natural language inference](https://mlsanity.com/paper/1508.05326/) (2015)
- [Large Language Models Encode Clinical Knowledge](https://mlsanity.com/paper/2212.13138/) (2022)
- [SuperGLUE: A Stickier Benchmark for General-Purpose Language Understanding Systems](https://mlsanity.com/paper/1905.00537/) (2019)
- [LaMDA: Language Models for Dialog Applications](https://mlsanity.com/paper/2201.08239/) (2022)
- [XNLI: Evaluating Cross-lingual Sentence Representations](https://mlsanity.com/paper/1809.05053/) (2018)
- [DROP: A Reading Comprehension Benchmark Requiring Discrete Reasoning Over Paragraphs](https://mlsanity.com/paper/1903.00161/) (2019)
- [Large Language Models Struggle to Learn Long-Tail Knowledge](https://mlsanity.com/paper/2211.08411/) (2022)
- [Are Transformers universal approximators of sequence-to-sequence functions?](https://mlsanity.com/paper/1912.10077/) (2019)
- [What BERT is not: Lessons from a new suite of psycholinguistic diagnostics for language models](https://mlsanity.com/paper/1907.13528/) (2019)
- [WinoGrande: An Adversarial Winograd Schema Challenge at Scale](https://mlsanity.com/paper/1907.10641/) (2019)
- [The Debate Over Understanding in AI's Large Language Models](https://mlsanity.com/paper/2210.13966/) (2022)
- [Shortcut Learning of Large Language Models in Natural Language Understanding](https://mlsanity.com/paper/2208.11857/) (2022)
- [Learning and Evaluating General Linguistic Intelligence](https://mlsanity.com/paper/1901.11373/) (2019)
- [Meaning without reference in large language models](https://mlsanity.com/paper/2208.02957/) (2022)
- [DALLE-2 is Seeing Double: Flaws in Word-to-Concept Mapping in Text2Image Models](https://mlsanity.com/paper/2210.10606/) (2022)
- [Why AI is Harder Than We Think](https://mlsanity.com/paper/2104.12871/) (2021)
- [DALL-E 2 Fails to Reliably Capture Common Syntactic Processes](https://mlsanity.com/paper/2210.12889/) (2022)
- [Language model acceptability judgements are not always robust to context](https://mlsanity.com/paper/2212.08979/) (2022)

## Built on by (most cited first, 30 of 45)

- [How Linguistics Learned to Stop Worrying and Love the Language Models](https://mlsanity.com/paper/2501.17047/) (2025)
- [Prompting is not a substitute for probability measurements in large language models](https://mlsanity.com/paper/2305.13264/) (2023)
- [A Sentence is Worth a Thousand Pictures: Can Large Language Models Understand Hum4n L4ngu4ge and the W0rld behind W0rds?](https://mlsanity.com/paper/2308.00109/) (2023)
- [Language in Vivo vs. in Silico: Size Matters but Larger Language Models Still Do Not Comprehend Language on a Par with Humans Due to Impenetrable Semantic Reference](https://mlsanity.com/paper/2404.14883/) (2024)
- [Potemkin Understanding in Large Language Models](https://mlsanity.com/paper/2506.21521/) (2025)
- [Order Matters: Investigate the Position Bias in Multi-constraint Instruction Following](https://mlsanity.com/paper/2502.17204/) (2025)
- [AI, write an essay for me: A large-scale comparison of human-written versus ChatGPT-generated essays](https://mlsanity.com/paper/2304.14276/) (2023)
- [Linguistic Generalizations are not Rules: Impacts on Evaluation of LMs](https://mlsanity.com/paper/2502.13195/) (2025)
- [Revealing emergent human-like conceptual representations from language prediction](https://mlsanity.com/paper/2501.12547/) (2025)
- [The Philosophical Foundations of Growing AI Like A Child](https://mlsanity.com/paper/2502.10742/) (2025)
- [Large Language Model probabilities cannot distinguish between possible and impossible language](https://mlsanity.com/paper/2509.15114/) (2025)
- [Post-training makes large language models less human-like](https://mlsanity.com/paper/2605.07632/) (2026)
- [Derivational Morphology Reveals Analogical Generalization in Large Language Models](https://mlsanity.com/paper/2411.07990/) (2024)
- [Benchmarking Abstract and Reasoning Abilities Through A Theoretical Perspective](https://mlsanity.com/paper/2505.23833/) (2025)
- [Beyond "Hallucinations": A Framework for Stable Human-AI Reasoning](https://mlsanity.com/paper/2510.14665/) (2025)
- [A Comparative Investigation of Compositional Syntax and Semantics in DALL-E 2](https://mlsanity.com/paper/2403.12294/) (2024)
- [Language Models at the Syntax-Semantics Interface: A Case Study of the Long-Distance Binding of Chinese Reflexive ziji](https://mlsanity.com/paper/2504.02116/) (2025)
- [Community size rather than grammatical complexity better predicts Large Language Model accuracy in a novel Wug Test](https://mlsanity.com/paper/2510.12463/) (2025)
- [Do You Get the Hint? Benchmarking LLMs on the Board Game Concept](https://mlsanity.com/paper/2510.13271/) (2025)
- [Non-native speakers of English or ChatGPT: Who thinks better?](https://mlsanity.com/paper/2412.00457/) (2024)
- [For GPT-4 as with Humans: Information Structure Predicts Acceptability of Long-Distance Dependencies](https://mlsanity.com/paper/2505.09005/) (2025)
- [Rethinking the Understanding Ability across LLMs through Mutual Information](https://mlsanity.com/paper/2505.23790/) (2025)
- [MinosEval: Distinguishing Factoid and Non-Factoid for Tailored Open-Ended QA Evaluation with LLMs](https://mlsanity.com/paper/2506.15215/) (2025)
- [A comprehensive study of LLM-based argument classification: from LLAMA through GPT-4o to Deepseek-R1](https://mlsanity.com/paper/2507.08621/) (2025)
- [Representational Similarity Alignment of Figurative Language in Humans and Large Language Models](https://mlsanity.com/paper/2601.09041/) (2026)
- [Multilingual Large Language Models do not comprehend all natural languages to equal degrees](https://mlsanity.com/paper/2602.20065/) (2026)
- [Glamour muscles: why having a body is not what it means to be embodied](https://mlsanity.com/paper/2307.08598/) (2023)
- [What is a word?](https://mlsanity.com/paper/2402.12605/) (2024)
- [To Drop or Not to Drop? Predicting Argument Ellipsis Judgments: A Case Study in Japanese](https://mlsanity.com/paper/2404.11315/) (2024)
- [Digestion Algorithm in Hierarchical Symbolic Forests: A Fast Text Normalization Algorithm and Semantic Parsing Framework for Specific Scenarios and Lightweight Deployment](https://mlsanity.com/paper/2412.14054/) (2024)

## Abstract (arXiv)

Large Language Models (LLMs) are recruited in applications that span from clinical assistance and legal support to question answering and education. Their success in specialized tasks has led to the claim that they possess human-like linguistic capabilities related to compositional understanding and reasoning. Yet, reverse-engineering is bound by Moravec's Paradox, according to which easy skills are hard. We systematically assess 7 state-of-the-art models on a novel benchmark. Models answered a series of comprehension questions, each prompted multiple times in two settings, permitting one-word or open-length replies. Each question targets a short text featuring high-frequency linguistic constructions. To establish a baseline for achieving human-like performance, we tested 400 humans on the same prompts. Based on a dataset of n=26,680 datapoints, we discovered that LLMs perform at chance accuracy and waver considerably in their answers. Quantitatively, the tested models are outperformed by humans, and qualitatively their answers showcase distinctly non-human errors in language understanding. We interpret this evidence as suggesting that, despite their usefulness in various tasks, current AI models fall short of understanding language in a way that matches humans, and we argue that this may be due to their lack of a compositional operator for regulating grammatical and semantic information.