Reference Hallucination in AI-Assisted Restorative Dentistry Education: An Educational Risk Assessment Using the Reference Hallucination Score Framework

Purpose: Artificial intelligence (AI) chatbots are increasingly used by dental students and faculty for literature review, coursework preparation, and scholarly writing. In restorative dentistry education, where evidence-based decision-making is fundamental, inaccurate AI-generated citations may pose educational risks. This study evaluated the bibliographic reliability of AI-generated references in restorative dentistry topics and examined their potential implications for dental education. Methods: Five large la…

By the Truvace deskOctober 15, 20261 min read1 sourcePeer-reviewedTRV-2026-1415· cite & verify

Reference Hallucination in AI-Assisted Restorative Dentistry Education: An Educational Risk Assessment Using the Reference Hallucination Score Framework — "09-8018-23" by NavyMedicine is marked with Public Domain Mark 1.0. To view the terms, visit https://creativecommons.org/publicdomain/mark/1.0/.

In brief

Researchers tested five large language model chatbots with ten standardised restorative dentistry prompts. Each chatbot generated 100 references, giving 500 in total. The team checked every reference by hand against indexed databases and scored it on an 11-point Reference Hallucination Score.

Dental students and faculty increasingly use chatbots for literature review and scholarly writing, so fabricated or garbled citations could weaken evidence-based training. The study reports that accuracy differed significantly between systems. The abstract gives no information on how often students actually rely on such citations or on whether verification habits reduce the harm.

Main points

Five chatbots each produced 100 references, for 500 references in total, which were verified manually against indexed databases.
ChatGPT-5.2 and Perplexity AI showed the lowest hallucination severity, while Copilot and DeepSeek-v3.2 showed the highest.
Prompt complexity did not significantly influence hallucination severity.
The most frequent errors involved author names, journal titles, and DOIs.

The problem

AI chatbots produced inaccurate bibliographic references on restorative dentistry topics, which can undermine evidence-based learning and academic integrity among dental students and faculty.

The rundown

The study used the Reference Hallucination Score, which ranges from 0 to 11, to rate each reference. Chatbots were compared with the Friedman test and Dunn-Bonferroni post hoc pairwise comparisons.

Binary error parameters were analysed with generalised linear mixed models. Chatbot type, prompt complexity and their interaction were fixed effects, and prompt was a random intercept.

The overall difference between chatbots was significant (p < 0.001), and specific model pairs differed significantly. The authors conclude that systematic reference verification and AI literacy are important in dental education.