AI in education
The future of education is being shaped by AI, but are we asking the right questions about its impact in the field? The recent [paper] called ““Don’t Forget the Teachers”: Towards an Educator-Centered Understanding of Harms from Large Language Models in Education” (https://arxiv.org/pdf/2502.14592) sheds some light on a perception gap problem we may have.
EdTech, a vast field including learning management systems (LMS), educational games, language learning apps, hardware for classrooms and remote learning, assistive technologies, and digital tools for educators, the paper suggests, faces a challenge:
currently, EdTech often prioritises technical concerns like toxic content and privacy, while educators are concerned about the wider impact on student learning, social development, and equity. Beyond just biased outputs, educators are focused on how LLMs influence critical thinking, collaboration, and the overall learning experience.
To create truly beneficial educational technologies, EdTech must:
- adopt an educator-centered design approach and co-design practices
- ensure that AI in education genuinely serves the needs of educators and learners
- design tools in a way that facilitates educator mediation of LLM harms
Additionally, regulators should develop centralised, clear, and independent reviews of LLM-based edtech, and create searchable repositories of vetted edtech tools. Besides, educator-centered procurement practices can significantly help bridge the gap between EdTech development and the actual needs and concerns of educators and learners.
There is a number of limitation this paper has, though: a very small and not representative sample of interviewees (they are all based in the US, the UK, and Canada).
The authors justly note: well-documented harms of and inequities in LLMs are not equally distributed and were not surfaced by the edtech providers and educators they spoke to. LLMs often:
- display cultural biases
- perform worse on low-resource languages
- are developed and operated with the labor and environmental costs not equally distributed
Nevertheless, it is a good starting point, and I hope there will be more research that look at this issue globally.
How can we bridge this divide, if it really exists? Educators, parents, and those with an interest in education, what broader impacts of AI do you see?
