A suite of LMs comprehend puzzle statements as well as humans

• URL: http://arxiv.org/abs/2505.08996v1
• Date: Tue, 13 May 2025 22:18:51 GMT
• Title: A suite of LMs comprehend puzzle statements as well as humans
• Authors: Adele E Goldberg, Supantho Rakshit, Jennifer Hu, Kyle Mahowald,
• Abstract summary: We report a preregistered study comparing human responses in two conditions: one allowed rereading, and one that restricted rereading.<n>Human accuracy dropped significantly when rereading was restricted, falling below that of Falcon-180B-Chat and GPT-4.<n>Results suggest shared pragmatic sensitivities rather than model-specific deficits.
• Score: 13.386647125288516
• License: http://creativecommons.org/licenses/by/4.0/
• Abstract: Recent claims suggest that large language models (LMs) underperform humans in comprehending minimally complex English statements (Dentella et al., 2024). Here, we revisit those findings and argue that human performance was overestimated, while LLM abilities were underestimated. Using the same stimuli, we report a preregistered study comparing human responses in two conditions: one allowed rereading (replicating the original study), and one that restricted rereading (a more naturalistic comprehension test). Human accuracy dropped significantly when rereading was restricted (73%), falling below that of Falcon-180B-Chat (76%) and GPT-4 (81%). The newer GPT-o1 model achieves perfect accuracy. Results further show that both humans and models are disproportionately challenged by queries involving potentially reciprocal actions (e.g., kissing), suggesting shared pragmatic sensitivities rather than model-specific deficits. Additional analyses using Llama-2-70B log probabilities, a recoding of open-ended model responses, and grammaticality ratings of other sentences reveal systematic underestimation of model performance. We find that GPT-4o can align with either naive or expert grammaticality judgments, depending on prompt framing. These findings underscore the need for more careful experimental design and coding practices in LLM evaluation, and they challenge the assumption that current models are inherently weaker than humans at language comprehension.

Related papers

• ExpliCa: Evaluating Explicit Causal Reasoning in Large Language Models [75.05436691700572]
  We introduce ExpliCa, a new dataset for evaluating Large Language Models (LLMs) in explicit causal reasoning.<n>We tested seven commercial and open-source LLMs on ExpliCa through prompting and perplexity-based metrics.<n>Surprisingly, models tend to confound temporal relations with causal ones, and their performance is also strongly influenced by the linguistic order of the events.
  arXiv  Detail & Related papers  (2025-02-21T14:23:14Z)
• One Thousand and One Pairs: A "novel" challenge for long-context language models [56.60667988954638]
  NoCha is a dataset of 1,001 pairs of true and false claims about 67 fictional books. Our annotators confirm that the largest share of pairs in NoCha require global reasoning over the entire book to verify. On average, models perform much better on pairs that require only sentence-level retrieval vs. global reasoning.
  arXiv  Detail & Related papers  (2024-06-24T02:03:57Z)
• Language in Vivo vs. in Silico: Size Matters but Larger Language Models Still Do Not Comprehend Language on a Par with Humans [1.8434042562191815]
  This work investigates the role of model scaling in determining whether differences between humans and models are amenable to model size. We test three Large Language Models (LLMs) on a grammaticality judgment task featuring anaphora, center embedding, comparatives, and negative polarity. We find that humans are overall less accurate than ChatGPT-4 (76% vs. 80% accuracy, respectively), but that this is due to ChatGPT-4 outperforming humans only in one task condition, namely on grammatical sentences.
  arXiv  Detail & Related papers  (2024-04-23T10:09:46Z)
• Data-Efficient Alignment of Large Language Models with Human Feedback Through Natural Language [31.0723480021355]
  We investigate data efficiency of modeling human feedback that is in natural language. We fine-tune an open-source LLM, e.g., Falcon-40B-Instruct, on a relatively small amount of human feedback in natural language. We show that this model is able to improve the quality of responses from even some of the strongest LLMs.
  arXiv  Detail & Related papers  (2023-11-24T15:20:36Z)
• Fine-tuning Language Models for Factuality [96.5203774943198]
  Large pre-trained language models (LLMs) have led to their widespread use, sometimes even as a replacement for traditional search engines. Yet language models are prone to making convincing but factually inaccurate claims, often referred to as 'hallucinations' In this work, we fine-tune language models to be more factual, without human labeling.
  arXiv  Detail & Related papers  (2023-11-14T18:59:15Z)
• SOUL: Towards Sentiment and Opinion Understanding of Language [96.74878032417054]
  We propose a new task called Sentiment and Opinion Understanding of Language (SOUL) SOUL aims to evaluate sentiment understanding through two subtasks: Review (RC) and Justification Generation (JG)
  arXiv  Detail & Related papers  (2023-10-27T06:48:48Z)
• Testing AI on language comprehension tasks reveals insensitivity to underlying meaning [3.335047764053173]
  Large Language Models (LLMs) are recruited in applications that span from clinical assistance and legal support to question answering and education. 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.
  arXiv  Detail & Related papers  (2023-02-23T20:18:52Z)
• Benchmarking Large Language Models for News Summarization [79.37850439866938]
  Large language models (LLMs) have shown promise for automatic summarization but the reasons behind their successes are poorly understood. We find instruction tuning, and not model size, is the key to the LLM's zero-shot summarization capability.
  arXiv  Detail & Related papers  (2023-01-31T18:46:19Z)
• Few-shot Instruction Prompts for Pretrained Language Models to Detect Social Biases [55.45617404586874]
  We propose a few-shot instruction-based method for prompting pre-trained language models (LMs) We show that large LMs can detect different types of fine-grained biases with similar and sometimes superior accuracy to fine-tuned models.
  arXiv  Detail & Related papers  (2021-12-15T04:19:52Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.