Newsletter - October 2024
Experiments with Gen AI
Gen AI is the new kid on the block and its a run away success in terms of usage, at least. How well its used, by whom, and for what purpose, remains to be seen. There are multiple ethical questions about what AI means for the human future, including how AI sees learning designers like myself (!!!!!), but here I’m more interested in its potential use for learning design.
In September, I made an effort to set aside some time to experiment with some new Gen AI tools. The first of these was Articulate 360, which launched its slightly delayed AI Assistant in September. This was offered on a 1-month trial (which is actually more like three weeks) to existing subscribers. I like some of the features, but the assistant doesn’t offer anything new in terms of Gen AI capabilities, is behind the curve in afew places, and has a few annoyances. Nonetheless, I’m sure the assistant will improve and it is perfectly adjusted to the needs of the Articulate platforms. Basically, it’s no game changer, but super useful if you are already engaged in the Articulate infrastructure for creating learning courses and activities.
The second, is Claude.ai, a chatbot that is a bit similar to ChatGPT, but in my opinion, able to produce some superior responses when I asked it to outline elements of an online course. What makes Claude particularly useful is the ability to upload articles and other sources for it to base its responses on. This is extremely useful for course building, but also has some useful applications for research!
I hadn’t even heard about the third AI chatbot that I experimented with until half way through the month, but it’s already looking to be a good one. This is Google’s NotebookLM. Like Claude, you can upload your own sources, but what makes this chatbot different is that it can generate a short (pretend) podcast where two AI voices chat about the contents of the source. This might sound strange (and it really is!), but it was surprisingly engaging and informative. It was also a much more enjoyable way to consume the contents of dense academic literature.
In this month’s newsletter, I propose to report back, briefly, on my findings. I hope that this is useful and if you do have any comments or thoughts of your own, please click on the comment box and leave a message!
Articulate 360 AI Assistant
I’ve tried to create a few courses using Articulate 360 AI Assistant while I had access to the trial. None of them quite meet my expectations, but in some ways that’s a good thing. What the AI Assistant is good at is providing a framework in which to build, and then allowing an expert to tailor and improve (and add detail and creativity). The AI Assistant is again useful for enhancing some of the text and has a super useful ability to create a summary for each unit of content (which is a pretty decent summary), and quiz questions (some of which are good, some need adjusting).
There are lots of different ways to use the AI Assistant and these differ for RISE and Storyline. For RISE you can upload a series of texts (articles, etc) which the AI Assistant will then use to suggest a outline for your course. This is useful, but has a few bugs. For starters it’s not always clear when you have exceeded the limits for the article files. You are only allowed so many it seems, but it doesn’t tell you when you have too many. This is particularly annoying as the learning outcomes, title, etc that you give the course disappears entirely if the AI Assistant deems that you have too many articles!
Aside from that problem, the outlines can be useful, but a little repetitive. They are always worth adjusting. When you get to the content the AI Assistant suggests topics related to the section heading. From these you can create most (but not all) of the content types in RISE. Again, these always need customising (as they also are repetative), but it’s useful for a starting point. For the paragraph option there is a useful option to provide additional instructions if the original paragraph is not to your liking. That includes increasing or decreasing the length and changing the style. If you change your mind about the type of content that you use, the AI Assistant also lets you switch - which is useful.
The ability to generate AI images isn’t really my thing. I find most AI images to be a bit ugly and fake, so would rather use my own, but I gave it a go and it does let you choose different styles and offers four options for each request, which is nice.
I’ve had less experience using the AI Assistant for Storyline, mainly because I hadn’t realised that it had already rolled out until late in the trial. That will therefore take more digging. For now, I’ve purchased the AI Assistant while it’s on a relatively cheep offer so that I can experiment more, but what happens when my renewal comes along I’m not sure. Articulate has bumped up the prices quite a bit this year and I don’t currently have any clients to use it with. I’m currently on the fence if this is useful enough for my own needs! We shall see.
Claude.ai
Claude is very similar to ChatGPT, but I find it more useful. Like the Articulate AI Assistant you can upload a series of articles into its database so that it can use these in its responses. I asked Claude to create a course outline for me and it did. While there were changes that I would make, the outline seemed pretty decent and Claude offered some suggestions and details which actually made me think more about what that course might be and how it could be improved.
What is useful with Claude is that you can easily separate discussions, so Claude essentially can continue a conversation at any time without other conversations that you might have with it, getting in the way. This is useful if you want to come back later and ask Claude to suggest some customisation or to delve deeper into one of the sections. I asked Claude to write me a transcript for a video I could create as part of the course, and it gave a pretty good outline, including some suggestions for ‘settings’ and imagery to use.
In one experiment I fed Claude a few chapters from an e-book about learning design and asked it to create a course outline for me. This is what is responded:
Introduction to Learning Design: Online Course Outline
Course Description
This course introduces students to the fundamental principles and practices of learning design. Participants will explore various instructional design models, learning theories, and practical techniques for creating effective learning experiences in diverse educational settings.
Course Objectives
By the end of this course, students will be able to:
Understand key principles of learning design and instructional theory
Apply various instructional design models to create learning experiences
Analyze learning needs and develop appropriate learning objectives
Design and develop effective learning materials and assessments
Evaluate the effectiveness of learning interventions
Module 1: Foundations of Learning Design
What is learning design?
Historical context and evolution of the field
Key learning theories and their implications for design
Module 2: Instructional Design Models
ADDIE model
Backward Design
SAM (Successive Approximation Model)
Design Thinking in learning design
Module 3: Needs Analysis and Learning Objectives
Conducting needs assessments
Analyzing learner characteristics
Writing effective learning objectives
Bloom's Taxonomy and its application
Module 4: Designing Learning Experiences
Structuring content and sequencing
Choosing appropriate instructional strategies
Creating engaging learning activities
Incorporating multimedia elements
Module 5: Assessment and Evaluation
Types of assessments (formative and summative)
Designing effective assessment items
Rubrics and scoring guides
Evaluating learning interventions
Module 6: Technology in Learning Design
Overview of learning technologies and platforms
Selecting appropriate tools for different learning contexts
Accessibility and universal design for learning
Emerging trends in educational technology
Module 7: Practical Application
Case studies in various educational settings
Developing a sample learning design project
Peer review and feedback
Reflection on the design process
Final Project
Students will create a comprehensive learning design project that demonstrates their understanding and application of the course concepts.
I asked NotebookLM to do the same, and I preferred Claude’s response. Nonetheless, using multiple AI chatbots and designing something that uses a bit from each is probably more effective.
I would certainly use Claude again as a ‘thinking board’, but whether I would use it to help me to build a course, I’m not entirely sure. I enjoy the process of creation too much to hand that job over to AI. But perhaps it can be a useful sounding board?
NotebookLM
NoetbookLM is similar to Claude AI in how it functions. You can upload a series of articles and ask the AI questions about it. Again I asked it to create a course outline for me from the content and it did that pretty well. It works by letting you create notes (which you can save). These seem useful especially as it provides references which can take you back to the original texts. However, when you save the note these links seem to disappear (which is a bit frustrating).
In one experiment I fed NotebookLM a few chapters from an e-book about learning design and asked it what a good learning design might be. This is what is responded:
Here is a discussion about what constitutes good learning design, drawing on the provided sources.
The sources provided focus on different learning theories and types of learning. They do not explicitly define “good learning design.” However, they offer valuable insights into the nature of learning and the various factors that contribute to effective learning experiences. Based on these insights, one could extrapolate some key principles that could be considered for “good learning design.”
Principles of Good Learning Design
Learner-Centered Approach: Good learning design should prioritize the learner's experience and needs. It should acknowledge that learners actively construct their own understanding based on their prior experiences, knowledge, and perspectives. Andragogy, as discussed by Knowles, emphasizes the importance of learner-centered approaches that utilize learners' past experiences.
Experiential Learning: Learning design should provide opportunities for learners to engage in experiential learning. This involves creating situations where learners can interact with the real world or simulated environments, allowing them to learn through direct experience and reflection.
Social and Collaborative Learning: The sources emphasize the social nature of learning. Good learning design should facilitate interaction and collaboration among learners, recognizing that they learn from and with each other. Vygotsky, for instance, highlights the value of collaboration and the Zone of Proximal Development, where learners can achieve more with support from others.
Reflective Practice: Learning design should encourage learners to reflect on their experiences and to critically analyze the information they receive. Brookfield stresses the importance of critical thinking, which involves challenging assumptions, exploring alternatives, and engaging in reflective skepticism. Similarly, Mezirow's concept of transformative learning highlights the process of critically reflecting on meaning perspectives and potentially transforming them.
Contextual Relevance: Effective learning design should connect with the learners' social, cultural, and socioeconomic contexts. This means understanding the learners' backgrounds, needs, and the environments in which they will apply their learning.
Multiple Types of Learning: Good learning design should cater to different types of learning, including cognitive, physical, emotional, and social learning.
Addressing Diverse Learning Needs
It is important to remember that "good" learning design is not one-size-fits-all. It needs to be tailored to the specific learning objectives, the target audience, and the context in which the learning will occur. The sources illustrate that there is no single "best" theory or type of learning. Therefore, good learning design should:
Consider the developmental stage of the learner. Piaget's stages of cognitive development, while subject to critiques, point to the need to align learning content with the cognitive capabilities of the learner.
Accommodate different learning styles and preferences. Marton and Säljö's work on surface and deep learning highlights how individuals approach learning tasks differently.
Recognize the social purposes of learning, which may be individual or collective.
By incorporating these principles, learning design can create engaging, meaningful, and effective learning experiences that promote individual growth and contribute to positive social change.
Please note that the definition of "good learning design" is subjective and can vary depending on the specific learning context and goals. The information presented here is based on an interpretation of the sources provided and may not encompass all perspectives on this topic. You may want to consult additional resources and engage in further research to explore a broader range of views on good learning design.
As this response is based on peer-reviewed sources it’s actually quite good and usable. If I was doing it properly I would have been more careful to select sources from a range of authors (not just one). It was also longer than I expected (although I could have asked NotebookLM to reduce the amount of text if I wanted to.
The key hook for NotebookLM though is its ability to create a short podcast, between two American sounding AI hosts. It’s surprisingly effective and on a first listen engaging. However, listen to a few of these in short succession and I bet they begin to sound a bit repetitive. Nonetheless, these audio elements could be useful for course builds to help students to understand a subject (if used only occasionally).
When I had finished writing this newsletter post I fed it into NotebookLM. Here is the podcast that resulted from it. I’m cringing massively when I listen to this, but that’s probably because it’s regurgitating my words back to me!!! I’m not sure it quite gets my hesitance about how useful the tools are. Do I mention anywhere that they will save me time? I don’t think I do, although the implication is perhaps there. I certainly do not consider this to be my “AI Brain-storming partner”. There is defiantly a tendency to over-dramatise reality going on here. In addition, the NotebookLM seemed to have a preference for (no guesses here) NotebookLM!! Anyway, I’d be interested in what you think of this?
Final Thoughts
I’m still not sure if or how I will use these AI tools in the future. It’s more about playing in the sandbox and seeing what happens. They are all potentially useful for learning designers, but there is definitely an application here for students to use these tools as well, as part of their learning (and thus there is an urgent need for new study skills training).
There is also potential here for researchers. All three of these new Gen-AI systems allow you to upload specific sources. That’s super-useful as it provides the opportunity to embed rigour into the process. Of course, Articulate is the odd-one-out here, as its AI Assistant is designed specifically for course design, which makes it less useful for direct learning or research, but perhaps more useful for course and activity design than the other two.
Claude and NotebookLM have many potential uses, but are, obviously less attuned to specific needs. So, it’s most likely a balancing act, and perhaps all three tools have different strengths which could be used in tandem.