PulseAugur
EN
LIVE 13:40:56

AI models tested on admitting ignorance, Claude hedges, Qwen3.8 Max errs confidently

By PulseAugur Editorial · [1 sources] ·

A recent test evaluated 12 AI models on their ability to admit when they don't know an answer, rather than confidently providing incorrect information. The study found that Qwen3.8 Max was the most prone to confident errors, while Claude models like Sonnet, Haiku, and Opus tended to hedge their answers, even when correct. None of the models could correctly answer a question about the probate application fee in England and Wales, with most defaulting to an outdated fee. AI

IMPACT Highlights the need for AI models to accurately express uncertainty, impacting future development of more reliable AI systems.

RANK_REASON The item describes a novel evaluation methodology for AI models and presents findings from that evaluation. [lever_c_demoted from research: ic=1 ai=1.0]

Read on dev.to — LLM tag →

AI-generated summary · Google Gemini · from 1 sources. How we write summaries →

AI models tested on admitting ignorance, Claude hedges, Qwen3.8 Max errs confidently

How we ranked this

Signal score
39 / 100
Composite score across the factors below. Higher = stronger signal that this story matters right now.
Newsworthiness bucket
Tool
The item describes a novel evaluation methodology for AI models and presents findings from that evaluation. [lever_c_demoted from research: ic=1 ai=1.0]
Source corroboration
Single-source cluster
Only one publisher covered this so far. Single-source stories can still rank when the publisher is high-authority, but they lack cross-source corroboration.
Topics
product, other
Editorial topic classification. Feeds into how the story surfaces on /topic/<slug> hub pages and into the per-entity coverage mix.
AI-industry relevance
High
Clearly on-topic for AI-industry coverage.
Story freshness
Breaking (< 6h)
Fresh story with cross-source coverage still developing. Ranking may shift as more sources report.

Full methodology in our editorial standards.

COVERAGE [1]

  1. dev.to — LLM tag TIER_1 English(EN) · Ben Dixon ·

    I scored 12 AI models on whether they admit they don't know. My own token limit nearly broke the test.

    <p>Most AI leaderboards ask whether the model got it right. I wanted to know something else: when it's wrong, does it know?</p> <p>So I put 100 questions to 12 models, each question with one answer you can check against an official page. Two grader models from other labs marked e…

RELATED ENTITIES

RELATED TOPICS