6 ways we’re using AI to develop online learning

As we navigate the future of education, we’re blending human expertise with AI innovations to improve the way we make online learning. Here are six ways we’ve used AI to inspire and assist us in our work.

1. Writing case studies for learning

MSc Project Management

Background: We faced a couple of challenges: we were starting work later than planned, and our academic colleague was new to teaching and had not developed any online learning before. They wanted to use activities based around case studies, but these can be labour-intensive to develop and require a great deal of creativity.

Solution: The learning designer and academic used generative AI to do the hard work. By crafting detailed prompts they were able to make ChatGPT generate bespoke case studies. They then developed questions and tasks for students to complete, based on the case studies.

Outcome: ChatGPT’s output has been surprisingly relevant and has helped us to structure our activities clearly; this was incredibly valuable for an academic with limited teaching experience, and it saved a significant amount of time during the development phase of the module. However, while we were getting started we spent a lot of time experimenting and learning how to generate effective prompts to produce the material we needed. And it was essential to sense-check the results to catch any errors or pieces of misinformation. Once you’ve worked that out, it is a powerful tool and easy to use!

2. New ideas for learning activities

BSc Business Administration

Background: Our academic experts sometimes encounter obstacles when brainstorming activity ideas, or are constrained by time during module planning. Our learning designers are used to suggesting ideas for activities but they are not subject experts and sometimes a very specific example is needed for inspiration.

Solution: Our learning designer turned to ChatGPT to generate several detailed ideas for activities. These activities were designed to align with the learning outcomes, and were then discussed with the academics and rigorously reviewed to ensure they were relevant, accurate and suitable for the module.

Outcome: The material produced by AI has been received positively. Some sceptical academics have been impressed with how relevant and detailed activities can be and we have found that the material can be used with only a few small edits.

3. An on-demand teaching assistant

Undergraduate Law degrees

Background: Our undergraduate and postgraduate law degree programmes are light on human interaction so we have been seeking ways to boost student engagement and interaction, and we wanted to explore how AI could help.

Solution: We’ve been piloting an AI teaching assistant that students can call on when they need help. Content from our degrees has been uploaded to the assistant which means that students now can engage in conversational interactions about the content of their programme. In addition, the software can also be used to generate formative assessments, learning activities and even provide assessment feedback. If required, the software can be extended to use the full large language model that supports ChatGPT4.

Outcome: The pilot will conclude late 2023 and feedback will be communicated soon after this.

4. Writing formative assessment questions

MSc Cyber Security

Background: Multiple choice questions (MCQs) are a good way for students to gauge their progress and understanding, but can be time consuming and laborious to produce.

Solution: Learning designers, learning technologists and academics worked together find a way to get AI to help. We took the scripts from the module’s video lectures, uploaded them to ChatGPT and prompted it to generate MCQs along with feedback for each correct and incorrect answer. The quizzes were then checked by the academics before upload to the platform.

Outcome: AI generated relevant questions to test understanding of the lecture content. Some of the feedback it generated was limited and required additional input from the academics. However, despite this, the quiz generation saved significant amounts of time, and the process can now be used in future.

5. Producing interactive activities

MSc Data Science

Background: A content-heavy module needed updating and we wanted to give students much more opportunity to practise and receive feedback on their understanding.

Solution: The learning designer and learning technologist took advantage of some of the new features of H5P to automatically create new, more interactive activities. Using the H5P Smart Import tool, existing video and written content was converted automatically into the following enhanced resources:

  • in-video questions
  • instructive text
  • a glossary
  • formative quizzes
  • flash cards
  • a summary

The output was checked by academics and then exported and embedded alongside the other learning materials in the online module.

Outcome: The initiative is still in development and feedback from the academic has been positive so far.

6. Formatting quiz questions for upload

BSc Business Administration

Background: This programme frequently uses multiple choice questions in the learning activities and they are all in a standard format. It is a challenge to format the questions for upload and doing this manually is time consuming and can be prone to error.

Solution: We used ChatGPT to format the questions for us: ChatGPT was given detailed instructions on the required structure of the format for quizzes, then the output was generated in correct format for uploading directly to the platform.

Outcome: It took a while to get things working properly: the prompt to generate the MCQs in the correct format for import to the platform had to be adjusted a number of times before it worked successfully. However once working the results have been excellent. It has saved a significant amount of time and, importantly, reduced opportunities for human error.

Other AI initiatives

Summaries for lectures in BSc Computer Science: Generative AI is used to generate short, easily digestible summaries to complicated video lectures.

Improved scaffolding for labs in MSc Cyber Security: ChatGPT is used to generate ‘hints and tips’ at the end of lab sessions to provide additional scaffolding and support for students.

Guidance for sensible use of generative AI: We have formed a group called ‘AI tools for module development’ which meets to discuss and evaluate the many AI tools and developments available. The plan is to provide detailed advice and guidance to help when using generative AI to develop academic content. This will form part of our new Module Development Handbook – a reference guide for our academics.

More efficient video editing: The video team has been trialling the use of new AI text-based editing functionality in Premiere, which allows you to edit a video using the video transcript.

Identifying trends in student feedback data: The module enhancement team has been using AI tools to analyse the open comments from the end-of-module student feedback surveys. AI helps identify, collate and categorise trends in the comments and undertake high-level sentiment analysis. All synthetically generated output can be anonymised and depersonalised before being reviewed by the team.

Images in this post were created with Adobe Firefly.