Discover more from The Learning Science Newsletter, Powered by DOMS™️
3 Game-Changing Use Cases for AI in Education
How I used AI tech this week to flip my classroom, 10X the speed of my research & run deep user analyses
AI is rapidly becoming part of my day to day work as an educator.
Here are three use cases which show some of these changes in action and the impact they’ve had on my effectiveness and efficiency (so far!).
Three Key Takeaways
On Efficiency: even the most basic, free-to-use AI like ChatGPT has huge potential to ~10X educator efficiency.
On Effectiveness: AI needs you! Without the right prompts + contextual information, AI is lost. In order for it to positively impact our effectiveness as educators, we need to provide the right prompts & parameters.
On Getting AI-Ready: to derive value from AI, we educators need to do two things. First, we need to understand *enough* about how AI works and its associated strengths and risks. Second, we need to develop a deep understanding of pedagogy + domain expertise. Together, these things will optimise our ability to prompt & critique AI in a way that could transform how we work. Without it, AI is more of hindrance than a help.
Use Case 1: Teaching Assistance & Classroom Flipping
Before AI: Participants on my Course Creator Bootcamp would do some analysis of their learners’ motivations using the Learner Transformation Framework™️ (LTF). Once they’d profiled their learners, they would spend ~4 hrs per week for 4 weeks working through a pedagogically and strategically complex process of turning the Discovery data into robust decisions about their course topic, format, content, activity, feedback & interaction.
As non-specialists, this part of the experience was more painful than it was delightful; my learners’ goal was to design and deliver the best possible, most high-yielding course possible as quickly as possible. Developing their learning design knowledge and skills was not on their to-do list.
AI came to their rescue.
After AI: Using ChatGPT + prompts provided by me, participants are now able to work with AI to complete their LTFs in <10 minutes. With the right prompts, they are able to turn this structured data into a set of pedagogically robust and market-optimised recommendations on course topics, objectives, format, content, activity, feedback & interactions in seconds.
Using ChatGPT + smart prompts, learners are able to generate independently what used to require hours of coaching & instruction.
Using AI to support learners to generate data and make decisions that I previously taught them means that I have been able to “flip” the classroom and dedicate the valuable time I have with my learners for analysis, evaluation and iteration of outputs - rather than spending time walking through the basics.
Using ChatGPT also means that I can deliver more value and more outcomes to my learners. In v1 of the course we only had the time to create a course outline. Powered by AI, v2 of the course enables learners to design, build, market & validate their courses in the same period of time.
It’s early days so my data is limited, but initial insights suggest that the impact of using AI on learner motivation, satisfaction & achievement has been pretty staggering.
TL;DR - my value proposition and impact is significantly more compelling than it was without AI.
Aside: the next cohort of the Course Creator Bootcamp will kick off in February. There are only a couple of spots left, bu if you want to take part you can find more info and apply for a place here. Join us!
Use Case 2: Research & Abstract Writing
I’m speaking at a lot of conferences in 2023, which is great. It also means lots of time spent on ideas generation and abstract writing.
Before AI: For each conference paper I presented, I would spend ~30 mins reviewing details of the conference purpose, fellow speakers’ profiles and my speaker’s brief. From there, I would spend ~90 mins coming up with an idea, course title and abstract to submit to organisers in line with with the brief. I’d then spend ~10 minutes formatting and checking the submission, before writing an email and sending it over.
After AI: Here’s how I rapidly generated ideas and wrote an abstract in ~10 mins this week:
First, I copied and pasted the conference agenda, information on fellow speakers and any email correspondence I’d had about the conference into ChatGPT. I gave ChatGPT the context and asked it to summarise the data into bullet points, highlighting specific things like the conference purpose and desired outcomes. It did this perfectly, in seconds.
Then, I asked ChatGPT to recommend one piece of research from each of my three fellow contributors to help me to understand the context of their papers. Again, it did this perfectly in seconds.
Using Elicit, I was able to quickly access short summaries of each piece of research and ask clarifying questions to quickly get a clear understanding of their research themes and purpose. TL;DR - in under 5 minutes I was more prepared than I have ever been to write a title and abstract.
But it didn’t end there. Next, I pasted my bio from LinkedIn and my website along with some writing samples into ChatGPT, at each step explaining what I was showing it.
Then, I asked ChatGPT asked it to suggest some ideas for a conference paper based on a) my research / expertise, b) my approach / style as shown in my wiring samples, c) the conference brief and c) the likely contributions of my fellow speakers. The results were impressive: ChatGPT generated an impressive range of suggestions in line with my research interests and style, in seconds.
After a couple of edits there was one more job to do. I asked ChatGPT to suggest a title and a generate a 100 word abstract based on the data I’d inputted. I asked it to include a reference to how my abstract related to the work of other contributors and instructed it to use an academic style and format but also to reflect the style of my existing work.
Finally, I asked ChatGPT to generate me an email to conference organiser thanking them for their invitation and sharing my abstract and bio. Job done.
ChatGPT + smart prompts has made me ~10X more efficient at generating & communicating ideas.
ChatGPT is intelligent *enough* to be able to respond to some relatively niche requests. E.g. it understands the structures and semantics of academic paper writing, which means that when I ask for something to be academic in style and tone (like an abstract for an academic conference), it can deliver.
Use Case 3: Data Analysis
I’m currently working on a project with the brilliant team at BBC Maestro to think about what online learning might look like in the future.
As with any project, it’s important that I gather as much context as possible to understand design-critical data, e.g.
User data: who Maestro users are, their age, location, what motivates them to learn, what they love, what would delight them even more etc.
Company data: the org vision, target markets, growth data & strategy, marketing strategies etc.
Before AI: On a project of this size, the sheer volume of data was both a blessing and a curse. In most projects, time and budget constraints would inevitably mean that decisions had to be made about what sorts of data to prioritise and what data to leave in the filing cabinet. TL;DR - before AI, the Discovery phase of the project was rarely optimised to include all data, which inevitably introduced risk to our decision-making.
After AI: AI loves data. This week, with some help from Chat GPT, I was able to review, summarise and structure hundreds of pieces of both qualitative and quantitative data in <10 minutes.
This included asking ChatGPT to summarise dense data like organisational strategy and budgets and summarising it into bullets or exporting it to Excel in a defined structure. Excitingly, based on user data, I was also able to ask ChatGPT to create a series of learner personas based.
Example Prompt: Create a series of learner personas from this data using the follow format: “I am X. I am Y years old and live in Z. I bought Y course in Z month because I wanted to X. Before purchasing this course, I had already tried X and Z, but these strategies failed because Y.
I left satisfied because the course delivered Z value. The experience could have met my needs even better if X.”
Once ChatGPT had sufficient contextual data, I could - through "Chain of Thought Prompting” ask it to do a range of incredibly powerful things, including:
Suggest a high-value course topic & format for someone of X profile.
Suggest the best marketing mode for X profile type, e.g. blog or video
Generate marketing content to catch the attention of X profile (and recommend where to post it).
With the right prompts & questioning, AI’s ability to review, summarise & restructure data in seconds (e.g. it can create tables or export data to Excel / Sheets) means thats it’s possible to conduct deep, robust Discovery more quickly and easily than ever before.
AI makes it possible for non experts to analyse larger datasets than ever before. This has significant positive impact on our ability to design learner-centred learning experiences optimised for motivation and value.
A final note on the risks of AI. One thing that’s very clear from my experimentation is that, in order to derive value from AI, we educators need to do two things.
First, we need to understand *enough* about how AI works in order to understand both its strengths and risks.
Second, we need to develop a deep understanding of pedagogy + domain expertise. Only armed with this knowledge can we can build the right prompts and critique AI’s outputs in a way that delivers more, not less, value to our learners.
What else can I try? Has AI impacted your day to day work? I’d love to hear from you in the comments below, or over on LinkedIn (give me a follow for more stuff like this!).
If you want to learn more, you can sign up for my AI for Educators course and join me and educators like you on a journey into the world of AI and Education (coming soon).
You can also check out my website, where you’ll find free guides & information on my courses, communities and subscriptions.