The Revolution Has Arrived: What the Current State of Large Language Models in Education Implies for the Future

TL;DR

This paper reviews the rapid integration of large language models into education since 2022, discussing their impacts, challenges, and future design considerations to ensure effective and accepted educational technologies.

Contribution

It provides a comprehensive overview of LLM applications in education, analyzes their successes and failures, and offers insights into future design challenges and paradigms.

Findings

LLMs have significantly impacted educational approaches since 2022.

New interaction paradigms with LLMs are becoming ubiquitous.

Design considerations are crucial for future acceptance of educational AI.

Abstract

Large language Models have only been widely available since 2022 and yet in less than three years have had a significant impact on approaches to education and educational technology. Here we review the domains in which they have been used, and discuss a variety of use cases, their successes and failures. We then progress to discussing how this is changing the dynamic for learners and educators, consider the main design challenges facing LLMs if they are to become truly helpful and effective as educational systems, and reflect on the learning paradigms they support. We make clear that the new interaction paradigms they bring are significant and argue that this approach will become so ubiquitous it will become the default way in which we interact with technologies, and revolutionise what people expect from computer systems in general. This leads us to present some specific and significant considerations for the design of educational technology in the future that are likely to be needed to ensure acceptance by the changing expectations of learners and users.