¿Qué funciona y qué falla en las evaluaciones de aprendizaje mediante IA?
PODCAST INVITADO: Serie EdTechnical 3, Episodio 5
Escritores
Esta temporada, la financiadora de EdTech, Libby Hills, y el investigador de IA, Owen Henkel, continúan hablando con destacados investigadores, profesionales y educadores sobre el tema. Serie de podcasts de EdTechnical Se abordará la vanguardia de la IA en la educación. Se desglosarán conceptos complejos de IA en explicaciones no técnicas para comprender mejor las investigaciones y ayudar a los educadores a discernir la información útil entre la exageración generada por la IA.
“I don’t think that hallucinations are going to be solved, you have to take them seriously.”
Daisy Christodoulou
This is a milestone moment for EdTechnical. Libby and Owen are not just releasing a new episode, but a new chapter. EdTechnical is evolving from a podcast into a broader platform focused on applied research, strategic investment, and sharing evidence-based insights about AI in education.
In this episode Libby and Owen are joined by Daisy Christodoulou MBE, EdTechnical’s muy first guest from two years ago. Daisy is Director of Education at No More Marking and a leading voice in assessment. Daisy, Owen and Libby reflect over what’s changed in the two years since that first episode, including Daisy’s own views about the opportunities for AI use in assessment. Daisy shares what her team has learned through their recent experiments with AI work, including how falling model costs are unlocking new possibilities, and why human-in-the-loop systems are essential.
They also talk about the risks of direct AI feedback, the limits of voice-based assessment, and the persistent challenge of hallucinations. What’s working, what’s broken, and what’s worth building next in automated assessment?
Enlaces
EdTechnical website
Making Good Progress?: The future of Assessment for Learning paperback by Daisy Christodoulou
No More Marking
Escritores