Prompt Engineering for Assessments

Generative Artificial Intelligence is a great technology. It has revolutionised the way people work and study. However, there are issues around students cheating and teachers struggling. There are also problems with the integrity of examinations. Is it time to stop struggling against the inevitable and integrate A.I. into the classroom?

Schools are struggling with Artificial Intelligence. The teachers don’t know how to manage cheating; students are becoming dependent on it, and national assessment authorities are struggling to keep up. There are multiple suggestions on how to counteract A.I. However, it is challenging for organisations to keep pace with a technology that is evolving at such a rapid rate. So, instead of creating strategies to counteract it, why not develop methods for integrating it?

Education is based on strong traditions. Study the material. Submit the Assignment. Wait for the Results. However, with A.I. becoming a cornerstone of education, we have to rethink how students are assessed. Instead of grading students on work produced by A.I. maybe it’s time to start grading students on the quality of the ‘prompts’ they use to generate their work.

A.I. may seem simple; just type in your idea (prompt), and as if by magic, a narrative appears. However, looking closer, all is not as it seems. Similar prompts can generate varying results. Prompts using a certain turn of phrase or even varieties in lexicon can lead to vastly different results. There is also the issue of A.I. generating intelligent, thoughtful essays and sourcing amazing papers and books that perfectly align with the author’s intent. But, on closer inspection, the sources do not exist in reality. Not only has the A.I created a narrative, but it has also created fictional journals, papers and books to support it.

When talking to A.I., users must be very specific. Think of it as being like an online search. Typing in the word ‘sweatshirt’ will get thousands of results, but to get the desired product, details such as brand, colour, and detailing have to be entered. This is similar to Prompt Engineering.

Prompt Engineering is a method of developing the prompt entered into a Generative A.I. application. For example, if a student wanted to write an essay on Fifteenth Century Kings, they may type the prompt ‘Create an essay about Fifteenth Century Kings’ However, the result will be very long and clumsy. Also, it lacks details such as class, year, and language. The prompt needs to be developed. Each time they add new details to the prompt, they will see the results improve.

What started with an initial prompt of ‘Create an essay about Fifteenth Century Kings’ could be engineered into ‘Create a 1500-word essay on Fifteenth Century English Kings, using three examples. The language should be suitable for a 13-year-old student with an Upper-Intermediate level of English. It reflects a middle-high school level of understanding of society and uses contemporary terms and language. Sources quoted should be on the GCSE curriculum or secondary/supplemental reading list.’

There is quite a difference between the two prompts. One could even say a ‘gradable’ difference. If a student just used the first prompt, they clearly didn’t put much thought into what they were asking for. However, the second prompt shows that the student is self-aware, knows what is expected of them and put a lot of thought into their prompt.

Instead of making teachers filter through pages and pages of possibly A.I. generated work, students can be graded on how good their Prompt Engineering skills are. Teaching these skills will engage the students more. By developing their prompts, students are learning to develop their ideas and think about what they are trying to say and how they are trying to say it. It also creates a situation where students may become invested in their work, as they developed the prompts themselves.

The author

Dr. John Jennings

Dr. John Jennings is an educational theorist from Galway, he has a PhD in TESOL with research in Social-Media and Education. He has researched the perceptions that students have of Social-Media in Education and how it affects their interaction with the academic world. He is also an avid virtual runner

https://linktr.ee/dr.johnjennings