‘Wow, you got way better results than I did’.
A recent comment from a colleague about the use of Artificial Intelligence got me thinking about the importance of our prompts and to analyse just how different prompts can produce vastly different results.
When it comes to teaching and taking on new tools or approaches, I am always drawn to the ‘input’ and ‘output’ of the process. I believe that any significant ‘input’ on the behalf of the teacher must result in significant ‘output’ for our students. Putting huge amounts of time and effort into various initiatives or programmes without significant benefit for the students makes little sense in my opinion. Some approaches are light on ‘input’ and heavy on ‘output’ (e.g. self-correcting assessments), whereas others are heavy on ‘input’ and light on ‘output’ (e.g. constructing hardcopy displays for standalone topics that need recreating weekly).
However, when it comes to AI prompting I have found that the ‘output’ will almost always match the ‘input’. The level of detail in your prompt is an indicator of how detailed and bespoke the suggestions will be. I have included an example below to demonstrate this: on the image above you will see a prompt for ‘creating a reader’s theatre script for a class of ten year old children on the topic of chocolate’. The second prompt (below) was to ‘create a reader’s theatre script for a class of 16 ten year olds on the theme of chocolate, where every character has at least 3 speaking roles’.
The initial prompt does not have enough speaking roles for the amount of children in the class, and all speakers have one speaking role – however, the revised prompt ensures it is a text that the entire class can look at and provides multiple speaking parts for each child.
Prompting in AI is so important in order to maximise its potential in education. The more specific and detailed we can be with our prompts, the more bespoke the ‘output’ will be. Sourcing materials that are suited to your own individual contexts is something that was extremely difficult in the pre ‘Gen AI’ age – now, careful and specific prompting can bring this about in a matter of seconds.