Teachers Talk Radio AI, Assessment and History teaching: The Late Show with Tom Rogers
8 snips
Jun 12, 2025 Daisy Christodoulou, an assessment expert focusing on History teaching and associated with No More Marking, joins Tom for a compelling discourse on the intersection of AI and educational assessment. They critically examine flawed mark schemes and explore the potential of AI tools in revolutionizing essay evaluations. Tom shares insights from his own experiments with AI and discusses the necessity of human judgment in assessments. The chat is a must-listen for anyone invested in the future of educational evaluation and innovative teaching practices.
AI Snips
Chapters
Transcript
Episode notes
Tom's AI Marking Experiment
- Tom Rogers experimented with AI marking a Year 8 history assessment on the British Civil War.
- The AI made repeated errors in simple tasks and missed nuances in essay analysis, requiring Tom to re-mark manually.
Essay Marking Reliability Limits
- Marking essays has inherent unreliability; markers disagree with each other and even themselves.
- Ofqual studies show a typical margin of error around ±5 marks in 40-mark essays, even under exam board conditions.
Comparative Judgment Improves Reliability
- Comparative judgment reduces marking errors by asking judges to compare pairs rather than make absolute judgments.
- Multiple judgments from many judges help cancel individual errors, improving reliability compared to traditional marking.