QTI-Certified Authoring, Now with AI-Generated Items

Assessible’s Content Editor already handles high-stakes exam authoring at scale. Together with the Institut für Bildungsevaluation (IBE), Trifork Amsterdam has added AI item generation. Built into the Editor itself, this new feature comes with fact-checking and human review as standard parts of the process to ensure high-quality assessment items.

The Content Editor is a QTI-3 certified authoring platform containing unlimited item banks, 15+ interaction types, item version control, shared stimulus and media banks, and author-defined metadata to support psychometrics and exam construction.

The next step was including AI. Because the Editor supports high-stakes exams, it was clear that generated items have to be held to the same standard of quality as any manually built items. This is why Trifork Amsterdam partnered with IBE, who built their AIG capability with machine learning experts at EPFL Lausanne and psychometricians at the University of Zurich. It produces mathematics and reading comprehension items natively in English, German, and French, with fact-checking built into the service and human review on top.

What Trifork built on their side is the authoring experience. Choosing a topic, setting parameters, and generating items all happen inside the editor. Authors review and edit generated items just like any manually built ones, so human feedback still majorly shapes the result.

Finished items go straight into an item bank as ordinary items — same tools, same review and approval, same assessments — or download as a standards-compliant QTI package for an external delivery engine.

More on the technical side of the generator, including the modelling approach, the architecture, and the guardrails, is coming soon.

Find a full article of the collaboration between AIG and the Editor here: