Latest / The Test Community.Network
16th July: The Limitations of AQA's AI Authoring Research, with Tim Burnett
In this week's episode, Tim Burnett goes solo to dig into AQA's new research report, Evaluating AI‑Authored Assessment Items for High-Stakes Exams, and to challenge the way it approached generating items with AI. Drawing on his own research sprint and recent workshops with awarding organisations, he sets out what the report gets right, where he thinks it falls short, and what it really takes to author good assessment items with generative AI. They discuss:
Why testing AI "out of the box" misses the point — the success of AI item generation lives in the process (the prompts, the harness, the…
The skinny
The skinny isn't ready yet — notes appear once the transcript is processed.