Cash et al. (2025)

Quantifying uncert-AI-nty: Testing the accuracy of LLMs’ confidence judgments

Memory & Cognition, 54(2), 375-400

Find It

DOI

Abstract

Abstract The rise of Large Language Model (LLM) chatbots, such as ChatGPT and Gemini, has revolutionized how we access information. These LLMs can answer a wide array of questions on nearly any topic. When humans answer questions, especially difficult or uncertain questions, they often accompany their responses with metacognitive confidence judgments indicating their belief in their accuracy. LLMs are certainly capable of providing confidence judgments, but it is currently unclear how accurate these confidence judgments are. To fill this gap in the literature, the present studies investigate the capability of LLMs to quantify uncertainty through confidence judgments. We compare the absolute and relative accuracy of confidence judgments made by four LLMs (ChatGPT, Bard/Gemini, Sonnet, Haiku) and human participants in both domains of aleatory uncertainty—NFL predictions (Study 1; n = 502) and Oscar predictions (Study 2; n = 109)—and domains of epistemic uncertainty—Pictionary performance (Study 3; n = 164), Trivia questions (Study 4; n = 110), and questions about life at a university (Study 5; n = 110). We find several commonalities between LLMs and humans, such as achieving similar levels of absolute and relative metacognitive accuracy (although LLMs tend to be slightly more accurate on both dimensions). Like humans, we also find that LLMs tend to be overconfident. However, we find that, unlike humans, LLMs—especially ChatGPT and Gemini—often fail to adjust their confidence judgments based on past performance, highlighting a key metacognitive limitation.

Reference

Cash, T. N., Oppenheimer, D. M., Christie, S., & Devgan, M. (2025). Quantifying uncert-AI-nty: Testing the accuracy of LLMs’ confidence judgments. Memory & Cognition, 54(2), 375-400. https://doi.org/10.3758/s13421-025-01755-4

What Rests on This

2 mechanisms cite this source. Each finding below is quoted as it appears on that mechanism's own page.

Feeling of Knowing Metacognition

Measured in AI

Same-question human–AI studies exist only for confidence given after answering, which bears on retrospective confidence rather than FOK: on trivia, 110 people's ratings discriminated right from wrong answers at an area of 0.50 and four chatbots (each run 100 times) at 0.48 to 0.65

Confidence Calibration Metacognition

Measured in AI

A peer-reviewed five-study paper (online 2025, in print 2026) gave online US adults and public chatbots (ChatGPT, Bard/Gemini and, in Studies 4–5, Claude Sonnet and Haiku, 2023–2025 versions, each run repeatedly through its chat interface) identical items and confidence scales, but its items were NFL and Oscar forecasts, sketch guessing, 22 more-than/fewer-than trivia questions built to require piecing facts together (110 adults), and 33 questions about an unnamed university's unpublished records (110 adults), so no study isolates confidence in answers recalled from memory
Measured in AI

On the trivia questions the best system, Claude 3.5 Sonnet (released 2024), missed its own prospective score by 2.15 items versus 3.51 for people (SD 3.15), with AUROC .66 versus .55 (SD .10), while systems ranged from 2.15 to 4.94 items and .57 to .66; on university questions Claude 3.7 Sonnet (2025) missed by 1.08 items versus 4.84 (SD 3.74), yet people discriminated right from wrong answers better (.57 vs .53), and over all 24 calibration comparisons LLMs were better 13 times, equal 4 and worse 7