Setting the Scene: The Myth of the Perfect Human Marker
• The halo effect of academic expertise
• Variability in human scoring — even among experts

Enter AI: Not Just a Cheaper Marker, But a Different One
• Capabilities of current AI models in marking open-ended responses
• How natural language processing (NLP) and LLMs are reshaping feedback and scoring
• AI’s advantages: scalability, speed, and yes—even transparency

“Consistently Inconsistent” – And That’s a Feature, Not a Bug
• Understanding that AI’s weaknesses can be diagnosed and improved
• Comparing human bias vs algorithmic bias: who’s really more accountable?
• The idea of “auditable fallibility” in machines

Changing the Narrative: Humans as the Benchmark, or the Bottleneck?
• Challenging the assumption that human judgement is always the gold standard
• Reframing AI as a tool to augment, not replace
• Why it’s not about less human, but more honest about our limitations

Looking Forward: What Needs to Happen Next?
• Regulatory and ethical frameworks that acknowledge both strengths and shortcomings
• Transparency in AI models and decision-making
• Investing in digital assessment literacy—for institutions, educators, and learners