Student Assessment and Artificial Intelligence – Exploring What’s Possible

A three component post here, as I want to bring together three different elements related to a presentation I gave in Indianapolis, IN for the International Center for Academic Integrity Conference 2023. The presentation was on student assessment and artificial intelligence. This post includes a video version of the presentation (re-recorded due to technical difficulties live at the conference), the accompanying slides, and a summary of key points.

The background to this presentation may be interesting. I submitted the session proposal prior to the release of ChatGPT, intended to show how I’d worked with technologies, including the GPT-3 model, to produce answers in the style of those required from students. With ChatGPT, more people can do this without any great technical understanding, so I refocused the presentation a bit to focus on ChatGPT and to suggest ideas about how we could approach assessment in the future. As various other recent posts suggest, there is still no definitive approach that works best, and I believe that discussion will need to happen at individual institutions, at a subject level, and involving students.

The Video

For my most detailed thoughts about the topic, please do watch the video recording. I go into more depth here than in the version I presented at the conference (which ended up being rather truncated due to the technical challenges – alas you will not hear the laughter at my jokes).

The Slides

If you would prefer to scan through the slides, you can see them here on my SlideShare account.

The Summary

During the presentation, I also talk briefly about how to get ChatGPT to produce a summary of a presentation (I also show how I can produce a press release). I then asked ChatGPT to summarise the five main take-home messages. Here’s what I found out from ChatGPT.

  • AI can be used by students to generate content for assignments, but it can lead to wrong information and academic misconduct.
  • AI can be a starting point and a time-saver for students if they know how to program and structure the AI syst
  • Prompt engineering is a significant skill that students need to learn to use AI effectively.
  • There are four approaches universities can take in dealing with AI in academic settings: ignore, ban, embrace, and educate.
  • AI can be used to generate feedback for research papers or writing difficult emails, but it should not be used for assessment as it may not be appropriate.

I don’t know if these would be my personal take-home messages, but it’s interesting to see what resonates with an individual (or an AI)

If you’re interested in a slightly longer summary, you can find a tweet with more details here.

1 thought on “Student Assessment and Artificial Intelligence – Exploring What’s Possible”

  1. Dr Stephanie Bull

    Thank you for this presentation. Such an important topic for educators and very well presented.

    My take home messages were different to Chat GPTs and I thought that it may be useful to share.

    1) Chat GPT is a language model and not a maths model. It is important to recognise what it can do, and what it can’t do (yet).

    2)AI is going to be part of the world that students are graduating into. Hence, it can’t be ignored, or banned, rather we should educate about it and to embrace it’s use in the education we provide.

    3) Chat GPT can improve productivity, but foundational knowledge/skills are still required. These may be discipline specific e.g. coding or conceptual e.g. critical appraisal, academic integrity

Leave a Comment

Your email address will not be published. Required fields are marked *

This site uses Akismet to reduce spam. Learn how your comment data is processed.