Genitem (Jan et al., 2026) is a tool that combines several AI agents and the AI-GENIE R package (Russell-Lasalandra et al., 2026), along with human expertise, to generate items for questionnaires, surveys, and measurement scales, and to verify their metric quality. It operates in two stages.
First, a configurator coordinates the work of the AI agents operating within a hierarchical structure. The researcher uses this to define the AI model to be used and each agent’s role. For example, an initial agent may draft a set of items based on a given theme, while three to eight proofreading agents independently verify that the items adhere to the topic, follow drafting rules, and are free from ethical bias. This review cycle repeats several times, with each agent incorporating feedback from the others before a final agent compiles the full set into a pool of items. AI-GENIE then processes this pool, checks its psychometric consistency, and identifies and retains the most stable items.
GenItem’s strength lies in its ability to simplify the technical aspect, making it easy to understand, whilst allowing human intervention to play a primary role in directing the AI agents.
URL : https://genitem.recherche-formation.com
That’s such a brilliant idea! Well done, me…
Anonymous author.