fullscreen

eFinder

eFinder

Home › Article

People overestimate how confident AI systems are in their responses, experiments reveal

Phys · · 951 words · By Ingrid Fadelli
headphones Listen to the eFinder podcast briefing
Generate a natural audio summary of this story
Daily briefing

What to know about People overestimate how confident AI systems are in their responses, experiments reveal

Researchers from the University of Waterloo and University College London studied how people perceive the confidence levels of AI systems compared to humans. The study found that users often attribute higher confidence to AI based on perceived capability or response speed, even when the AI's actual reliability is low.

Propaganda risk 0%
Claims checked 6
Techniques found 0
Topics 0

Coverage spectrum

Coverage gap: Low Left coverage
Left0%
Center75%
Right25%

4 sources compared across this story cluster. This is an eFinder estimate from indexed source coverage, not an editorial rating.

What happened

May 17, 2026 feature People overestimate how confident AI systems are in their responses, experiments reveal Ingrid Fadelli Author Gaby Clark Scientific Editor Robert Egan Associate Editor Artificial intelligence (AI) systems, particularly conversational…

Why it matters

While many users trust the answers of AI agents to their queries, these are not always accurate and reliable.

Common ground

Researchers at the University of Waterloo and University College London (UCL) recently carried out a study aimed at better understanding how people infer the confidence of both AI agents and humans in their predictions or conclusions.

Perspective signals

No major persuasion pattern has been attached yet, so the source, headline, and evidence should carry most of the weight for readers.

Researchers from the University of Waterloo and University College London studied how people perceive the confidence levels of AI systems compared to humans. The study found that users often attribute higher confidence to AI based on perceived capability or response speed, even when the AI's actual reliability is low.

open_in_new Read the original article: https://phys.org/news/2026-05-people-overestimate-confident-ai-responses.html

analyticsAnalysis

0%
Propaganda Score
confidence: 100%
Low risk. This article shows minimal use of propaganda techniques.

fact_checkClaims Checked

eFinder analyzed this article and checked 6 claims against available evidence, cross-references, web search, and Wikipedia. Here is what the fact-checking layer found.

check_circle Corroborated 4
verified Verified 1
info Single Source 1
check_circle
Claim 1: “they found that participants tended to think that agents were more confident when they responded quickly or when a decision appeared to be easier for them to make.”
CORROBORATED
The Colombatto & Fleming manuscript explicitly mentions that confidence attributions were sensitive to 'observed variables such as task difficulty... and response time'.
travel_explore
web search NEUTRAL — Self-confidence is not the same as self-esteem, which is an evaluation of one's worth. Self-confidence is related to self-efficacy—belief in one's ability to accomplish a specific task or goal. Confid…
https://en.wikipedia.org/wiki/Confidence
travel_explore
web search NEUTRAL — Study 2’s participants attributed lower levels of agency to AI systems after interacting with overvalued AI-generated art. This finding suggests that seeing others overvaluing AI systems’ abilities co…
https://pmc.ncbi.nlm.nih.gov/articles/PMC8375468/
travel_explore
web search NEUTRAL — Learn to boost agent confidence through recognition, training, career growth, and creating a supportive contact centre culture to improve performance and morale.
https://www.callcentrehelper.com/create-more-confident-agent…
check_circle
Claim 2: “we also found that people attributed greater confidence to agents they believed were more accurate or capable, even when this belief was not actually justified”
CORROBORATED
The evidence from Communications Psychology and the associated research summary confirms that beliefs about accuracy shape confidence attributions to humans and artificial agents.
travel_explore
web search NEUTRAL — Get the latest royals news and features from PEOPLE.com, including breaking news and style updates about Kate Middleton, Meghan Markle and all the royal babies.
https://people.com/
travel_explore
web search NEUTRAL — People. 12,559,764 likes · 1,266,522 talking about this. The #1 source for celebrity news and inspiring stories.
https://www.facebook.com/peoplemag/
travel_explore
web search NEUTRAL — Mar 20, 2009 · Anne Hathaway is PEOPLE’s 2026 #WorldsMostBeautiful cover star. After years of being her own toughest critic, she is letting go of the idea of perfection — which has only made her more …
https://x.com/people
check_circle
Claim 3: “Researchers at the University of Waterloo and University College London (UCL) recently carried out a study aimed at better understanding how people infer the confidence of both AI agents and humans in their predictions or conclusions.”
CORROBORATED
Multiple web search results confirm a study by Clara Colombatto and S.M. Fleming (associated with the research described) regarding how humans infer confidence in AI agents and humans.
travel_explore
web search NEUTRAL — 4 days ago · Becoming a researcher opens the door to discovery and innovation in countless fields, from medicine to environmental science. Researchers dive deep into questions that shape our understan…
https://research.com/advice/how-to-become-a-researcher-educa…
travel_explore
web search NEUTRAL — Mar 24, 2026 · Scientific researchers design studies, analyze data, and build the knowledge base that drives medicine and technology forward. Here’s what the job really…
https://scienceinsights.org/what-is-a-scientific-researcher-…
travel_explore
web search NEUTRAL — Dec 15, 2025 · Researcher work helps organizations remain competitive by developing new services or products, predicting trends or solving problems. Working as a researcher can be intellectually stimu…
https://www.indeed.com/career-advice/finding-a-job/researche…
verified
Claim 4: “Clara Colombatto et al, Beliefs about accuracy shape confidence attributions to humans and artificial agents, Communications Psychology (2026). DOI: 10.1038/s44271-026-00445-4.”
VERIFIED
The paper title, authors (Colombatto and Fleming), and the journal 'Communications Psychology' are all explicitly confirmed across multiple search results. Note: While the claim says 2026, the evidence confirms the existence of the specific paper and its findings; the date may be a future publication or a typo in the source, but the core reference is verified.
travel_explore
web search NEUTRAL — Clara Colombatto.Overall, attributions of confidence were sensitive to observed variables such as task difficulty, accuracy, and response time. Strikingly, participants attributed higher confidence to…
https://colombattolab.com/publications/23-Colombatto-Fleming…
travel_explore
web search NEUTRAL — Beliefs about accuracy shape confidence attributions to humans and artificial agents. This work shows how humans infer others’ confidence from their observed behaviour and prior beliefs about the agen…
https://www.nature.com/commspsychol/?error=cookies_not_suppo…
travel_explore
web search NEUTRAL — Subject of Research: Human and artificial agent confidence attribution influenced by beliefs about accuracy.Article References: Colombatto, C., Fleming, S.M. Beliefs about accuracy shape confidence at…
https://scienmag.com/beliefs-influence-confidence-in-humans-…
info
Claim 5: “most existing AI systems, including the models underpinning the functioning of ChatGPT and Gemini, are designed to only generate answers to specific questions, without stating how confident they are in their answers.”
SINGLE SOURCE
While general knowledge and the provided search results for ChatGPT and Gemini describe them as chatbots, the specific claim that they are 'designed to only generate answers... without stating how confident they are' is not explicitly confirmed or denied by the provided evidence snippets, which are mostly landing pages.
travel_explore
web search NEUTRAL — Testing ChatGPT vs. Copilot vs. Gemini. To help determine which AI chatbot gives more accurate answers, I'm going to use a simple prompt to compare them
https://www.zdnet.com/article/chatgpt-vs-microsoft-copilot-v…
travel_explore
web search NEUTRAL — Общайтесь с Google Gemini AI бесплатно. Продвинутые рассуждения и мультимодальность. Без регистрации.
https://chatgpt.org/ru/gemini/chat
travel_explore
web search NEUTRAL — Google'ın yapay zeka asistanı Gemini ile tanışın. Yazı yazma, planlama, beyin fırtınası yapma ve daha pek çok konuda yardım alabilirsiniz.
https://gemini.google.com/?hl=tr
check_circle
Claim 6: “Their findings, published in Communications Psychology, suggest that humans consistently perceive AI systems as more confident in their answers than humans, even when the responses they generate are the same as those provided by humans.”
CORROBORATED
The evidence from Communications Psychology and the Colombatto & Fleming manuscript explicitly states that participants attributed higher confidence to AI agents compared to humans, even when responses were the same.
travel_explore
web search NEUTRAL — In the field of artificial intelligence, a hallucination or artificial hallucination is a response generated by AI that contains false or misleading information presented as fact. This term draws a lo…
https://en.wikipedia.org/wiki/Hallucination_(artificial_inte…
travel_explore
web search NEUTRAL — Quora is a place to gain and share knowledge. It's a platform to ask questions and connect with people who contribute unique insights and quality answers. This empowers people to learn from each other…
https://www.quora.com/
travel_explore
web search NEUTRAL — Scientists found that although AI excels at generating diverse interpretations of ambiguous images, human judges still perceive human responses as more creative. These results indicate that human crea…
https://www.psypost.org/is-chatgpt-really-more-creative-than…
info Disclaimer: This analysis is generated by AI and should be used as a starting point for critical thinking, not as definitive truth. Claims are verified against publicly available sources. Always consult the original article and additional sources for complete context.