Quanta research on source fidelity in AI cards

Explore Quanta's internal measurements of source binding in AI-generated study material, including methodology, raw data and clear limits on what each result demonstrates.

Published research

Source fidelity · June 2026

Do AI flashcards hallucinate?

Historical Google Gemini test from June 2026: all 1,997 delivered test flashcards carried an evidence quote found in the source. Of 2,042 generated candidates, 45 were discarded. The test has not been repeated with the current Mistral model. It measures source fidelity, not factual correctness or current product quality.

100% of delivered test flashcards source-backedRead

The evidence stays where you need it: on the card.

The study measures the same gate used before delivery in the Source-First path. Quanta retrieves real source text, generates from it and keeps the passage attached to the study content.

01You start with a topic or your own material.
02Quanta links every delivered Source-First card to its source and passage.
03If the evidence cannot be confirmed in that text, the candidate is dropped.

Confirmed evidence makes a statement traceable. It is not an automatic fact check.

Live search · Alpha-Zerfall

Serlo · „Radioaktivität"

Picked

Wikipedia · „Alphastrahlung"

Wikibooks · Physik

Quanta searches vetted, openly licensed knowledge sources for your topic and picks the best passage.

Study material you can back up

Every AI card with a source, or not at all. Start free.

Start for free