Think. Try. Transform. #2

Planning Spaced Repetition with AI

In the second edition of the Think. Try. Transform. series, I dive into the crucial interplay of memory, forgetting, and learning.

Think.

The German psychologist Hermann Ebbinghaus proposed his theory of the forgetting curve in the late 1800s. It has gone on to be a prominent theory in the science of learning due to the various implications he proposed:

• Forgetting is an exponential process initially, which eventually evens out over time

• The majority of forgetting occurs within the first 24 hours after a concept is learnt

• Stress and sleep are also important factors when it comes to memorisation

• Mnemonics and review techniques (such as spaced repetition) can mitigate against forgetting

The standard forgetting curve. Source: Wikipedia

The forgetting curve (red) with example of how the curve flattens with spaced repetition (green). Source: Wikipedia

How often do we take into account the forgetting that takes place in the classroom? I am well aware of many of the current methods to facilitate memorisation and other approaches, influenced by the work of Rosenshine’s Principles and Cognitive Load Theory for example. However, how specific do/should we get with understanding how much will be forgotten when we teach a new concept? Do we map out the additional learning that needs to occur in order for a concept to be deeply embedded in long-term memory? Or do we assume that regular retrieval exercises, spaced at varying lengths of time, is sufficient?

I know I certainly have not paid too much attention to this in the past. With the vast amount of content that needs to be covered, along with the other demands of my time within a school day, it is nigh on impossible to deeply consider this, for all areas of the curriculum being taught, in great detail. And perhaps we don’t have to. Perhaps we should start by focusing on several areas of new learning, that if embedded deeply in long term memory, have a profound impact on future learning. Starting here, I think, is more realistic. And it is in this endeavour that using AI can be incredibly powerful.

Try.

Using ChatGPT or equivalent, teachers can map out learning for particular concepts, with spaced repetition intervals. This can then be used in planning documents to ensure that key concepts are taught at intervals that promote retention in long term memory.

The key here, as with any interactions with any AI, is to ensure that you iterate towards the desired outcome. Below there is an example of prompt that you may decide to use, with various iterations, however, you may get somewhat different results from mine.

Prompt Example

Applying the principles of the forgetting curve, create a spaced repetition schedule for a # week block of work on the concept [NAME]. The students are in Year # and are taught the [NAME] Curriculum. Include specific days of the week to do the activities, assuming that learners make the expected progress across the # weeks.

Prompt: Applying the principles of the forgetting curve, create a spaced repetition schedule for a 6 week block of work on the concepts of metaphors. The students are in Year 6 and are taught the English National Curriculum. Include specific days of the week to do the activities, assuming that learners make the expected progress across the 6 weeks.

Iteration Examples

• Create a table of the spaced repetition schedule

• Ask the AI to brainstorm ideas for a particular activity

• Ask the AI to create assessment quizzes. Or even use Chris Goodall’s tips for creating a quiz which is then fed back into the AI for analysis.

• Reference the SuperMemo SM-2 algorithm in your prompt to generate a schedule based on a well known formula for spaced repetition. (The technically minded can ask the AI to generate VBA code which can then be extracted and used to create a spreadsheet that can track different pathways of spaced repetition based on additional metrics.)

Prompt: Applying the principles of the forgetting curve, create a spaced repetition schedule for a 6 week block of work on the concepts of metaphors. The students are in Year 6 and are taught the English National Curriculum. Include specific days of the week to do the activities, assuming that learners make the expected progress across the 6 weeks. Use the SuperMemo SP2 algorithm as the foundation for your decision making when designing the schedule.

The power in a use case such as this is the ability to extend the output in a myriad of different directions. With ChatGPT plus, you could upload curriculum documentation (you can do this for free with Claude) before the initial prompt; plot forgetting curves based on your repetition schedule; generate images for some of the tasks; create worksheets and other teaching resources.

Transform.

So, how can creating spaced repetition schedules transform your practice and the experience of the students?

• Enhanced memory retention: whilst imperfect, actually thinking about and planning for the forgetting students will experience, will go some way in ensuring that they retain more information in their long term memory.

• Tailored learning pathways: different pathways can be tailored for different individuals or cohorts. This is incredibly difficult to achieve consistently without the power of AI.

• Pedagogical prowess: quite simply, I think it will make better teachers.

• Data-driven decision making: more complex versions of a spaced repetition schedule, which use results or teacher judgements to guide the following activities, would mean teachers are making more informed, scientific decisions.

• Enhanced planning and organisation: it would add effective layers to a teacher’s planning, which could mean more efficient use of time and resources.

• Skill enhancement: using AI to perform a task such as this, particularly with the various iterations that result in more successful outputs, only enhances the tools a teacher has to do their job effectively.

Next week, I will take a look at how to use text to speech tools to create podcasts.