Language Models Fail to Introspect About Their Knowledge of Language

• URL: http://arxiv.org/abs/2503.07513v2
• Date: Wed, 12 Mar 2025 03:18:36 GMT
• Title: Language Models Fail to Introspect About Their Knowledge of Language
• Authors: Siyuan Song, Jennifer Hu, Kyle Mahowald,
• Abstract summary: We systematically investigate emergent introspection across 21 open-source language models.<n>We evaluate whether models' responses to metalinguistic prompts faithfully reflect their internal knowledge.<n>We propose a new measure of introspection: the degree to which a model's prompted responses predict its own string probabilities.
• Score: 13.743212705122751
• License: http://creativecommons.org/licenses/by/4.0/
• Abstract: There has been recent interest in whether large language models (LLMs) can introspect about their own internal states. Such abilities would make LLMs more interpretable, and also validate the use of standard introspective methods in linguistics to evaluate grammatical knowledge in models (e.g., asking "Is this sentence grammatical?"). We systematically investigate emergent introspection across 21 open-source LLMs, in two domains where introspection is of theoretical interest: grammatical knowledge and word prediction. Crucially, in both domains, a model's internal linguistic knowledge can be theoretically grounded in direct measurements of string probability. We then evaluate whether models' responses to metalinguistic prompts faithfully reflect their internal knowledge. We propose a new measure of introspection: the degree to which a model's prompted responses predict its own string probabilities, beyond what would be predicted by another model with nearly identical internal knowledge. While both metalinguistic prompting and probability comparisons lead to high task accuracy, we do not find evidence that LLMs have privileged "self-access". Our findings complicate recent results suggesting that models can introspect, and add new evidence to the argument that prompted responses should not be conflated with models' linguistic generalizations.

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. We tested seven commercial and open-source LLMs on ExpliCa through prompting and perplexity-based metrics. 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)
• Randomly Sampled Language Reasoning Problems Reveal Limits of LLMs [8.146860674148044]
  We attempt to measure models' language understanding capacity while circumventing the risk of dataset recall.<n>We parameterize large families of language tasks recognized by deterministic finite automata (DFAs)<n>We find that, even in the strikingly simple setting of 3-state DFAs, LLMs underperform un parameterized ngram models on both language recognition and synthesis tasks.
  arXiv  Detail & Related papers  (2025-01-06T07:57:51Z)
• Trustworthy Alignment of Retrieval-Augmented Large Language Models via Reinforcement Learning [84.94709351266557]
  We focus on the trustworthiness of language models with respect to retrieval augmentation. We deem that retrieval-augmented language models have the inherent capabilities of supplying response according to both contextual and parametric knowledge. Inspired by aligning language models with human preference, we take the first step towards aligning retrieval-augmented language models to a status where it responds relying merely on the external evidence.
  arXiv  Detail & Related papers  (2024-10-22T09:25:21Z)
• What Languages are Easy to Language-Model? A Perspective from Learning Probabilistic Regular Languages [78.1866280652834]
  Large language models (LM) are distributions over strings.<n>We investigate the learnability of regular LMs (RLMs) by RNN and Transformer LMs.<n>We find that the complexity of the RLM rank is strong and significant predictors of learnability for both RNNs and Transformers.
  arXiv  Detail & Related papers  (2024-06-06T17:34:24Z)
• Holmes: A Benchmark to Assess the Linguistic Competence of Language Models [59.627729608055006]
  We introduce Holmes, a new benchmark designed to assess language models (LMs) linguistic competence. We use computation-based probing to examine LMs' internal representations regarding distinct linguistic phenomena. As a result, we meet recent calls to disentangle LMs' linguistic competence from other cognitive abilities.
  arXiv  Detail & Related papers  (2024-04-29T17:58:36Z)
• LLMs' Reading Comprehension Is Affected by Parametric Knowledge and Struggles with Hypothetical Statements [59.71218039095155]
  Task of reading comprehension (RC) provides a primary means to assess language models' natural language understanding (NLU) capabilities. If the context aligns with the models' internal knowledge, it is hard to discern whether the models' answers stem from context comprehension or from internal information. To address this issue, we suggest to use RC on imaginary data, based on fictitious facts and entities.
  arXiv  Detail & Related papers  (2024-04-09T13:08:56Z)
• Prompting is not a substitute for probability measurements in large language models [22.790531588072245]
  We compare metalinguistic prompting and direct probability measurements as ways of measuring models' linguistic knowledge. Our findings suggest that negative results relying on metalinguistic prompts cannot be taken as conclusive evidence that an LLM lacks a particular linguistic generalization. Our results also highlight the value that is lost with the move to closed APIs where access to probability distributions is limited.
  arXiv  Detail & Related papers  (2023-05-22T17:33:17Z)
• Large Linguistic Models: Investigating LLMs' metalinguistic abilities [1.0923877073891446]
  We show that OpenAI's o1 vastly outperforms other models on tasks involving drawing syntactic trees and phonological generalization. We speculate that OpenAI o1's unique advantage over other models may result from the model's chain-of-thought mechanism.
  arXiv  Detail & Related papers  (2023-05-01T17:09:33Z)
• Large Language Models with Controllable Working Memory [64.71038763708161]
  Large language models (LLMs) have led to a series of breakthroughs in natural language processing (NLP) What further sets these models apart is the massive amounts of world knowledge they internalize during pretraining. How the model's world knowledge interacts with the factual information presented in the context remains under explored.
  arXiv  Detail & Related papers  (2022-11-09T18:58:29Z)

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.