The Need for Theory of AI in learning and teaching

Iterationism is specifically a theory of AI and learning. It looks beyond prompt engineering and points to a way of working in classrooms where students may have a shared task of improving AI-generated material for a specific purpose. We shall have more to say on the details of this but processes and products are iterated. They are iterated by the originators and by others and they can begin in class and be continued online in different modes. Iteration is the social glue in collaborative classroom and online learning supported by technology.

AI is advancing at pace with new tools appearing daily. The development of these technologies follows an iterative process as an acknowledged part of improvement and implementation. Although this is a central feature of technology-development, it is not so central in the way it is applied to learning. This is partly because we are still coming to terms with significant changes arising from conversational and generative tools. We are still trying to make sense of these technologies.

The struggle to make sense of these technologies has produced a number of traditional theories for learning which have then been applied to AI. These include Constructivism and Cognitive Load Theory. Such theories are equally useful with or without technology but Iterationism specifically recognises the qualities of the technology relative to ways of learning. Specifically, it is possible to demonstrate knowledge of a subject by showing your ability to improve it. Iterationism is not therefore about building knowledge in the way described in Constructivism, it is instead a way of exploring an issue, refining it, and achieving an improvement congruent with your understanding. This might be done individually, in small groups, or collectively as a whole class.

Iterationism follows the AI development process but in this theoretical framework, iteration is applied consciously and in different ways to learning, teaching, and research. It does this for individual and social learning. It applies equally for informal and formal education. Likewise, it applies as much for in-person or co-located education as it does for online and highly distributed collaboration. In all these cases, iteration with AI, is done multiple times as required and in order to achieve an individual or shared educational goal and to demonstrate understanding of the topic/subject.

Iteration is a feature of many models of learning, teaching, and research. It is central to Problem-Based Learning which uses iteration as a driver to solve a given task. Ideas are formed and developed as learners work towards a solution. Through repeated efforts, some attempt is made to solve or at least address a problem.

Action Research follows a similar principle as it journeys through several cycles in an attempt to firstly understand, and then secondly, to develop practice. This can often apply to a higher level of iteration. For instance, it may apply to a project-level iteration in which the practice itself is iterated.

In all these cases, iteration is feature rather than the core basis for learning design. This is distinct from Iterationism which sees iterations as an individual and shared approach which may be interconnected in purpose, process, and product.

In Iterationism, learning is designed to be organised around the use of iteration in response to a learning challenge. Class members may work collaboratively on how to iterate a given product. The outcomes of iteration then become a focus for sharing, evaluation, and discussion.

So, what does Iterationism mean? How does Iterationism differ and how does it relate to AI in Education? We argue that iteration is firstly ‘central’ to learning and teaching with AI. In other words, it is at the very heart of the approach and secondly, it is a way of thinking about learning, teaching, and research. This is true whether it is for a simple learning activity with AI tools or whether it is part of a larger collaborative or collective level of development. In either case, AI is used to generate material which is to be evaluated and improved. Iterationism is therefore a theory of improvement for the purpose of learning and teaching.

In other words, whereas iteration is a feature of other work such as PBL or Action Research, in Iterationism it is regarded as the engine of individual or social learning. The learner would have the idea of ‘iterating’ supported by AI or iterating the products of AI technology.

We recently ran a workshop on Iterationism and Collaborative Learning with AI for employees of MKC Training Services Ltd at the Royal School of Military Engineering at Chatham in Kent. Some of the feedback is shown below:

“There was some interesting content and discussion today….. It would be good to know of more creative ideas like this which could be used to take the lesson away from just being a teacher led experience.”

“Thank you for organising such an interesting CPD session 😊”

“…a very interactive and thought provoking session”

“It was very useful, enjoyable and interesting.”

Disclaimer: This site does not represent or express any views of the University of Derby. The University of Derby is not responsible in any way for any content or views expressed on this or any other CLAI Project site.