Webinar: Using AI to Generate Question Items PLUS New Guide Published

A Practical Guide to Authoring Test Items with Artificial IntelligenceTim Burnett, Founder of the Test Community Network, is publishing a brand-new guide: A Practical Guide to Authoring Test Items with Artificial Intelligence. In this live session, Tim will walk you through the guide, sharing the key findings, practical techniques, and insights gathered during a research sprint conducted with practitioners, psychometricians, and assessment professionals from across the international community.What the Guide CoversThe guide tackles one of the most pressing questions facing assessment organisations today: how do you use AI to generate high-quality multiple-choice items without compromising on validity, fairness, or security?Rather than treating AI as a magic button that produces ready-made questions, the guide sets out a structured, evidence-informed approach. It recognises that introducing AI into item authoring fundamentally changes the workflow — shifting the skill requirement from writing questions to instructing AI how to write questions well.What Tim Will Cover in the WebinarDuring the session, Tim will take you through the highlights of the guide, including:The two-stage prompt workflow — One of the guide's central recommendations is that item generation should not be treated as a single-step process. The first stage focuses on generating the question stem and correct answer, grounded in approved source material. Only once these have been reviewed and approved should distractors be generated in a second stage. This approach addresses a common pitfall: when stems, keys, and distractors are generated together, edits to one element can silently undermine the others.Different approaches to AI-assisted authoring — There is no single way to implement AI in your item development process. Tim will explore the practical trade-offs between using conversational chat assistants, configurable AI agents, and dedicated SaaS-based authoring platforms. Each comes with its own strengths and constraints, from the flexibility of building your own agent architecture to the structured workflows offered by embedded platforms.Tips and techniques from the research — Tim will share the practical lessons discovered during the sprint, including how to craft prompts that produce genuinely plausible distractors (not just random wrong answers), why uploading entire textbooks as context can do more harm than good, how to manage token costs without sacrificing quality, and why a "blind review" style SME answer check can prevent anchoring bias.Insights from the assessment community — The guide draws on contributions from practitioners across the sector. Tim will share some of the most valuable insights that emerged, from the challenges of keeping SMEs meaningfully engaged when AI drafts become increasingly polished, to the often-overlooked security risk of item pre-exposure when independent users generate similar items from similar prompts.