AI’s Impact on Instructional Design and Podcasting (Google's NotebookLM)

Sep 26

Sep 26 AI’s Impact on Instructional Design and Podcasting (Google's NotebookLM)

Somehow, it has become a part of my job to pay attention to the latest advancements with Generative AI. This is largely in part due to working at MIT and many of the new courses and programs my team has been developing revolves around Gen AI topics. With this in mind, it has made me extremely jaded by every new tool that comes about. There is a type of internal filtering system in my head that is filled with heavy skepticism on any tool before I use it and it takes quite a bit of effort to convince me otherwise.

Last week, one new tool was made available to the public that made me say whoa… That’s mind-blowing. Luckily, I found out about the tool from you, the learning nerds, and I appreciate it when you send me new tools to look into. The tool I’m talking about is Google’s NotebookLM. NotebookLM is an experimental AI-powered tool designed to help users understand complex information more efficiently. Think of it as a virtual research assistant that can summarize facts, explain complex ideas, and brainstorm new connections based on the sources you provide.

Here's how it works:

Simply upload your relevant documents, such as research papers, articles, or notes.
NotebookLM will analyze your sources and provide summaries, explanations, and answers to your questions.
Discover new connections, ideas, and perspectives based on the information in your sources.

I experimented with a few articles to see what NotebookLM produced and it was helpful, but GPT-4 still seemed similar to me. What I was really curious about was one feature in particular. NotebookLM can generate a podcast episode based on what you upload. Now, I know what you are thinking. Great, some clunky sounding robot voice is going to read my article back to me. Not so fast my friend. It’s a podcast with two hosts, a male and female sounding voice, and they banter back and forth about your material. NotebookLM even adds in breaths before they speak sometimes to make it sound more human! Who on earth thought of that?

Anyway, so for today’s podcast episode, I want you to listen to a sample of this. The article is around a chapter I wrote for a book about AI and instructional design. NotebookLM mistakenly thinks that I have a new book called AI and ID, and it refers to that in the beginning. This is only a chapter though. In here, I talk about Generative AI uses in the instructional design space with scenario-based learning, creating simulations, chatbots, personalized learning paths, role-playing, and analyzing feedback. I also speak a bit about the challenges of AI and the drawbacks.

So, let’s listen in to our AI friends discuss about the impact of AI on instructional design, and I’ll be back to share some thoughts:

Speaker 1
Remember that feeling of being hopelessly lost in a research rabbit hole? Sixth grade I was writing a paper on ancient Egypt, and I swear I lived at the library. Then a friend told me about Ask Jeeves and suddenly answers appeared like magic. These days it feels like AI is giving everyone those same Ask Jeeves vibes exciting. A bit mysterious.

Speaker 1
Maybe not as revolutionary as we think, though.

Speaker 2
It's like the metaverse all over again. Suddenly everyone's talking about it. But those of us familiar with virtual worlds were already living it in Second Life. Instructional design has this way of hyping up every shiny new thing, right?

Speaker 1
Virtual reality, gamification, microlearning. Yeah, but here's the thing. We're not about to tell you A.I. is the be-all and end-all solution for instructional designers. In this deep dive, we're going beyond the buzz to unpack a chapter from MIT expert Luke Hobson's upcoming book, AI ID.

Speaker 2
We'll break down exactly what A.I. is, the opportunities and challenges it presents for learning, and how instructional designers can apply these powerful tools thoughtfully. Think of this as your guide to cutting through the hype and getting down to practical applications.

Speaker 1
So before we get ahead of ourselves, let's define A.I. When we say A.I., what are we actually talking about?

Speaker 2
At its core, A.I. is about enabling machines to perform tasks that typically require human intelligence. We're talking problem solving, learning from data, even understanding natural language. On a technical level, it involves creating algorithms, essentially sets of rules that allow computers to mimic human cognitive processes.

Speaker 1
Okay, so we're teaching machines to think, but not necessarily in the same way our brains do. More like giving them the tools to analyze information and make decisions based on what they learn.

Speaker 2
Exactly. One key aspect of this is machine learning, a subset of A.I., where algorithms actually improve over time by processing and learning from data. They identify patterns, make predictions, and become more accurate the more data they're fed.

Speaker 1
This is where things clicked for me. I'll admit I was hesitant about A.I. at first, but then I tried using ChatGPT to help me draft learning objectives for a graduate course and it was like having a brainstorming partner, bouncing ideas back and forth.

Speaker 2
That's a perfect illustration of how A.I. can be a powerful tool for instructional designers. It's not about replacing us, but augmenting our abilities, helping us work smarter. Like Luke Hobson mentions in his book, it's a Kickstarter for ideas, but it still requires the human touch to truly shine.

Speaker 1
Speaking of shining, Hobson's chapter is full of examples of how A.I. is already being used in instructional design. One that caught my eye was using A.I. to develop realistic training scenarios. He talks about this case study revitalizing organizational culture at Tech Fusion Corp.

Speaker 2
That case study is a great example of how A.I. can add depth and nuance to scenarios. It's like having a team of writers helping you flesh out every detail, creating a much richer learning experience for participants.

Speaker 1
In the Tech Fusion example, A.I. crafted this whole narrative around a company struggling with low morale and disengagement. It outlines specific leadership approaches to address the problem, like listening tours and cultural ambassadors and even projected positive outcomes based on those strategies. It's incredibly detailed and really brings this scenario to life.

Speaker 2
That level of realism is crucial for effective learning. When we can immerse learners in authentic scenarios, they're far more likely to connect with the material and retain the information.

Speaker 1
Absolutely. And Hobson doesn't stop at scenarios. He talks about using GPT-4, a really powerful A.I. model, to create a simulation for aspiring instructional designers who are prepping for job interviews. Imagine if every professional could have access to that kind of realistic practice.

Speaker 2
The possibilities with AI-powered simulations are incredibly exciting. This example, in particular, highlights how A.I. can democratize learning, making these immersive, often expensive experiences accessible to a wider audience.

Speaker 1
He even used A.I. to brainstorm different breakout room activities for a workshop on adult learning.

Speaker 2
I bet it generated some creative ideas. It's easy to fall back on the same old activities, but A.I. can push us to think outside the box, offering fresh perspectives and challenging our assumptions.

Speaker 1
It really can. He talks about how helpful it was to use A.I.-generated suggestions for incorporating role-playing into a negotiation skills course.

Speaker 2
Role-playing is such a powerful tool for active learning, but creating those scripts can be time-consuming. A.I. can generate a solid starting point, which we can then refine and customize to align with specific learning objectives.

Speaker 1
So far, we've got scenarios, simulations, brainstorming, role-playing. Is there anything A.I. can't do? It's like the Swiss Army Knife of instructional design.

Speaker 2
Well, it can even help us repurpose transcripts, which are often overlooked but incredibly valuable resources.

Speaker 1
I mean, those things we typically just create for accessibility. Tell me more.

Speaker 2
Exactly. Transcripts are a goldmine of information. A.I. can help us extract key themes, generate discussion prompts for deeper engagement, and even create glossaries and knowledge checks to reinforce learning.

Speaker 1
Wow. I never thought of it that way. Transcripts become dynamic learning assets, not just static documents.

Speaker 2
Right. And we can go even further. A.I. can analyze learner data from various sources, identify patterns, and even personalize the learning journey in ways we could only dream of before.

Speaker 1
Okay, so let's say we have all this data about how learners interact with our content, how long they spend on each module, their quiz scores, which activities they engage with most. What happens then?

Speaker 2
A.I. can process and analyze that data much faster and more efficiently than humans. It can identify struggling learners early on, highlight areas where the instruction might not be effective, and suggest personalized learning paths to address individual needs.

Speaker 1
That's amazing. We're talking about a whole new level of personalized learning. All powered by A.I. It's like having a personal tutor for every single learner.

Speaker 2
It is. But before we get too carried away, we need to acknowledge the potential challenges of using A.I. in instructional design. It's not all sunshine and roses.

Speaker 1
You're right. Hobson talks about some of these challenges in his chapter. What are some of the big ones we need to be aware of?

Speaker 2
Well, one of the biggest is the question of human oversight. Just because we can automate certain tasks doesn't mean we can just set it and forget it.

Speaker 1
Right? We still need human instructional designers to make sure the technology is used effectively and ethically, and perhaps most importantly, to provide context and build those all-important human connections with learners.

Speaker 2
Exactly. Finding that balance between A.I.'s capabilities and the irreplaceable human element of learning is crucial. We'll dive deeper into that in the next part of our deep dive.

Speaker 1
We'll be back after a quick break to do just that. We're back and ready to keep going with A.I. and that human element in learning. Finding that balance you were talking about. How do we do that?

Speaker 2
It really is fascinating. We've always had this one-size-fits-all approach to learning, you know. And now with A.I., there's the potential to personalize things in ways we couldn't have imagined before.

Speaker 1
It's a big shift from that old factory model of education, that's for sure. More customized, more adaptable. But how do we design for that?

Speaker 2
That's the million-dollar question that really makes us rethink the entire design process. Instead of that fixed curriculum, we need to be designing these flexible learning ecosystems that can change and adapt to individual needs, individual goals.

Speaker 1
Now you're speaking my language. Yeah, but what does that actually look like? An adaptive learning ecosystem.

Speaker 2
So imagine a platform, right? And the A.I. is constantly monitoring the learner's progress, analyzing everything: how they interact, assessments, the feedback they give. And based on all that data, the system starts suggesting personalized activities, resources, even adjusting how fast or slow they go through it, even the difficulty level.

Speaker 1
So instead of waiting till the end of a module to realize someone's struggling, we can adjust things immediately.

Speaker 2
Exactly. And it's not just about catching people up either. Let's say a learner is doing really well, excelling. A.I. can identify that too and give them more challenging material or opportunities to dive deeper, creating that dynamic learning experience, always meeting the learner right where they are.

Speaker 1
That's incredible. It's like a personal guide for navigating the world of knowledge.

Speaker 2
I like that analogy. And it gets even more interesting. What if these adaptive learning ecosystems could also connect learners, you know, facilitate collaboration and knowledge sharing?

Speaker 1
So instead of everyone being in their own little personalized bubble, A.I. could connect them with others who have similar interests or learning styles, even similar goals.

Speaker 2
Exactly. Think about the possibilities for mentorship, connecting a learner with someone more experienced or even an expert in that field.

Speaker 1
Now, that's really something. We're talking about personalizing not just the learning experience but the whole learning network. But let's be realistic for a moment. How achievable is all of this? Are we talking about some distant future, or is this something we could actually see happening soon?

Speaker 2
That is the question, isn't it? The reality is we're already seeing pieces of it. Platforms like Duolingo, Khan Academy, they're already using A.I. to personalize learning pathways and give customized feedback.

Speaker 1
Right. And those chatbot tutors we talked about are getting more advanced all the time.

Speaker 2
Exactly. As A.I. technology keeps evolving, we can expect even more impressive applications. But of course, we need to acknowledge the challenges that come with all this, especially with the personalization and data analysis side of things.

Speaker 1
What kind of roadblocks should we be aware of as we move forward with A.I. and instructional design?

Speaker 2
Well, the biggest one is probably the sheer amount of data we're talking about. To create these truly adaptive systems, we need a lot of data—tracking, analyzing every single learner interaction, assessment, every piece of feedback. And we need to do all of that while respecting privacy and ensuring data security.

Speaker 1
Privacy is such a huge concern these days. Learners need to know their information is being handled responsibly.

Speaker 2
Absolutely. And it's not just about collecting the data. It's making sense of it all. We need A.I. algorithms that can actually analyze these massive amounts of information and translate that into useful insights to guide our design decisions.

Speaker 1
So we can't just throw A.I. at the problem and expect it to magically work. We still need to be thoughtful about how we're using it, make sure it aligns with what we're trying to achieve pedagogically.

Speaker 2
Precisely. And that brings up another point. We need a new kind of instructional designer—people who are experts in learning theory and pedagogy but also comfortable with data. They need to understand what A.I. can do, and what it can't do.

Speaker 1
A hybrid skill set, blending the art of teaching with the science of data and technology.

Speaker 2
Exactly. And those designers are going to be in high demand, especially as A.I. starts to reshape the field.

Speaker 1
It's exciting and a bit daunting at the same time, don't you think?

Speaker 2
It is, but I'm optimistic. You know, I think the potential for learners is huge, too.

Speaker 1
But before we get too lost in thinking about the future, let's ground this a bit. What are some concrete things designers can do today to start using A.I. in their work?

Speaker 2
Some really practical steps. It's a great question. Like I said, start small. Think about those examples from the book and don't be afraid to just dive in.

Speaker 1
Okay. Start small. What would that actually look like, though, for someone who's maybe just starting to explore A.I. in their work?

Speaker 2
Well, you know, you mentioned ChatGPT earlier. That's a great place to start, or even Google Bard, something like that. They can be really helpful for brainstorming, especially if you're feeling stuck—like try generating some dialogue for a role-playing activity.

Speaker 1
Oh, I like that.

Speaker 2
Yeah. Or you could even have A.I. help you analyze learner feedback.

Speaker 1
It's like having an extra brain in the room, one that can process information so much faster than we can.

Speaker 2
Exactly. And the thing is, these tools are constantly learning and improving. The more you use them, the more you figure out what they can do. It's a learning process for both you and the A.I.

Speaker 1
I love that. So it's not just us teaching the machines. We need to be open to learning from them, too.

Speaker 2
Yeah, absolutely. And that brings up another important point. We need to prioritize our own professional development in this area.

Speaker 1
Right, because things are changing so fast.

Speaker 2
Exactly. And there are so many great resources out there now—online courses, webinars, conferences—all focused on A.I. and education.

Speaker 1
So don't be intimidated by that learning curve.

Speaker 2
No, embrace it. This field is constantly changing. We've got to be willing to learn and adapt right along with it.

Speaker 1
It's like we're all students again, in a way. Yeah. Exploring this new frontier together. Which makes me think collaboration is going to be more important than ever. Don't you think?

Speaker 2
Oh, absolutely. Talk to your colleagues, share what you're learning—what's working, what's not. Go to conferences, join online communities. The more we all share, the better prepared we'll be as a field.

Speaker 1
I love that—collaboration over competition, especially when it comes to something like A.I. with so much potential to change things.

Speaker 2
I agree. At the end of the day, it's all about creating the best learning experiences we can. And if that means embracing new tools and new ways of thinking, well, then that's what we need to do.

Speaker 1
It's so easy to get caught up in all the excitement or even the anxiety around A.I. But as instructional designers, we've got to stay focused on using these technologies in a way that's thoughtful, ethical, always putting the learner first.

Speaker 2
Finding that balance—you know, making the most of what A.I. can do, but never losing sight of that human element that makes learning truly meaningful.

Speaker 1
Well said. It's been such a fascinating conversation. Thanks for joining me today to dive into the world of A.I. and instructional design. And for our listeners, if you have any thoughts, experiences, or even questions about A.I., please share them with us. Until next time.

So, what’s your first initial reaction? Are you surprised, alarmed, or turned off by the whole thing? It’s odd. I listened to about four of these AI generated podcast episodes and each one had their own quirks. There are definitely moments when you can tell it’s not human. The tone doesn’t quite match the level of enthusiasm where you would expect the voice to go up instead of down. There are a few breaks in speech that can seem a bit out of place at times. The oddest thing I experienced with this though is that after a while, you forget it’s AI. Glitches happen all the time when we are on Zoom calls so if someone sounds robotic or laggy, we blame the Wi-Fi. Is it really that much of a difference compared to listening to AI speak? Not really.

Another quirk is that the male voice went in and out of first and third person with the material. If you listened in the beginning, it sounds like he was taking the position of actually talking as me. The male voice referred to himself as being the ID and at times, like the author of the chapter. Later on though, he flips into a broader picture. This going in and out made it hard to follow along with at times.

Another glitch in the matrix was the content NotebookLM added to my content. For instance, when describing about learning pathways, I do describe how it’s valuable to place students together by cohort and even in mini-cohorts to match their interests. Nowhere in there do I refer to learning styles, but since AI loves learning styles, it had to throw that in the mix. I heard the same thing from a friend of mine who is a professor and weather scientist. His expertise is in studying lightning (cool, right?), and he mentioned the NotebookLM had thrown in several inaccurate statements. This has always been a problem with LLMs, and while it has gotten tremendously better in the last two years, it’s still something to be mindful of.

So, let’s talk a bit about the pros, the cons, and the what’s next.

For the pros, this is an unbelievable tool to study with. I have a million whitepapers saved on my laptop to read for later and guess who hasn’t read them yet? In theory, I can upload a whitepaper and have AI create a podcast for me on the research. As someone who loves to learn with educational podcasts, this is right up my alley and since you are listening to a podcast, I bet you will love this too. I tried it with a few samples, and I can absolutely see this being a new legitimate way to stay on top of current trends and issues. There are other features within NotebookLM that I am heavily glossing over with the note capability, coding efforts, collaboration tools, etc. There is more to NotebookLM than just the podcast generator, but that’s the one area I was interested in the most.

Now with the happy things out of the way, let’s talk about the cons. By far, the quality is still not there yet. With all the reasons I mentioned about a minute ago, I don’t see all of us now loving to only listen to AI generated content. For some of you, I’m sure you found it extremely hard to listen to and you didn’t enjoy that last segment of the show. As a podcaster, I’m certainly not going to complain about that! The misinformation is something I will always worry about and I could see that going downhill quickly. My fear, which I’m sure will happen soon if it hasn’t happened already, is that AI generated podcast are about to blow up. In 2023, we saw this when the public had access to ChatGPT and all of a sudden, Amazon was flooded with new books being co-authored with ChatGPT. Other publications had to shut down submissions because of the overwhelming amount of Gen AI produced material. Will this happen to podcasts? Oh, I’m sure it will.

For the what’s next, this technology is only going to improve. Remember when Siri came out and everyone was shocked that we could now “talk” to AI? ChatGPT has only kicked that concept up by several notches, and it sounds like Google is about to do the same. I would say to expect improvements and new ideas around this generated AI podcast space in the future. On the Brightside, you’ll still have the human version of me for this podcast for the foreseeable future.

What do you think? I’m curious what you have to say!

A huge thank you to our sponsors! Consider supporting those who support our show!