- Who it helps
- Educators and designers evaluating AI-assisted learning.
- Availability
- The October 22 preprint is public. This is a controlled study, not a released tutoring product.
A controlled learning experiment
Thirty-two participants studied science topics with assistants whose briefings contained different proportions of planted falsehoods. The study measured quiz answers and confidence.
Checking with the same chatbot
The authors report that learners often returned to the same chatbot to check answers. Independent source checks are a practical concern raised by the result.
Confident AI tutors can teach falsehoods with confidence ↗HCI Index week 43 featured cards ↗
What you can do
- Read the study design before applying its findings to a classroom or tutoring product. Open reference ↗
Where the evidence stops
The experiment uses a small sample and deliberately manipulated briefings. It does not estimate the error rate of every current tutor; results were not reproduced.
Sources read. Sources checked October 29, 2026. No installation, model inference or end-to-end workflow was executed.
Sources & further reading
Paper · paper
Confident AI tutors can teach falsehoods with confidence ↗
Establishes the subject, discovery context or stated verification limit.
arxiv.org · page checked 2026-10-29
Web · analysis
HCI Index week 43 featured cards ↗
Establishes the subject, discovery context or stated verification limit.
Prateek Solanki / HCI Index · page checked 2026-10-29
Coverage note: No qualifying current independent attention sample; original X posts unavailable. Reported outcomes were not independently reproduced.
The record over time (1)
Dated source checks and published explanations. An observation date can differ from the original release date.
- · Added to Current
A small experiment found that learners absorbed false facts from confident chatbots and remained confident in their answers.