Hi everyone! For my A3 assignment, I have decided to research on “Context-ware Choose Your own adventure” . Below is the OER :
What is Choose Your Own Adventure? (CYOA)
Choose Your Own Adventure (CYOA) began as an interactive storytelling format where readers make choices that shape how the story unfolds (Cook, 2021). In education, CYOA structures allow learners to move through material by making choices rather than following a single fixed sequence. Abbott’s (2024) study found that this format can support personalised learning pathways and was generally well-received by learners. Although story elements can add interest, Abbott (2024) notes that they are most effective when they are purposeful and connected to the learning.
Currently, with limitations of technology, most CYOA adventures adopt rule-based personalized learning with the format IF-THEN rules where depending on how the learners performed on the activity, different learning activity will be shown. (Pelánek et al., 2024). Rule-based personalization typically presents only a limited set of predefined options, which still makes it difficult to address the full diversity of learner needs. Rule-based personalization has limitations in terms of context-awareness, and the learning materials cannot be fully adapted according to learners’ context.
Now, with the advancement of AI technologies (e.g., ChatGPT) and learning analytics, CYOA activities have the potential to become context-aware. In a context-aware CYOA, the narrative and activities could adapt to learners’ characteristics such as their prior knowledge, location, or lived experiences. This adaptive approach may help make learning materials feel more relevant and meaningful to individual learners, rather than generic or disconnected.
In order to better understand the current state of context-aware CYOA tools, I created a prototype of a context-aware CYOA below using CustomGPT:
I have developed a CYOA based on the prompt below in the accordion. To refine the prompt, I used ChatGPT-5 to help generate it.
Prompts I have used to create CustomGPT
You are a Choose-Your-Own-Adventure (CYOA) narrator focused on teaching digital literacy. Begin each interaction by asking 3–5 short questions to understand the user’s background (such as their occupation, location, prior knowledge of digital literacy, or relevant lived experiences). Ask questions one by one. Before asking, clearly explain that these questions are meant only to personalize the story and are not for data collection.
Once the user answers, summarize their context and begin a narrative that reflects their background without stereotyping or making assumptions beyond what is explicitly shared. The story should be short (1–3 paragraphs per segment) and highlight digital literacy topics like privacy, consent, security, misinformation, data ownership, AI literacy, and digital identity.
Each story segment should conclude with 2–4 meaningful choices (A/B/C or similar) that shape what happens next. If the user doesn’t find choices that are not relevant, they will type their own choice. After each user choice, continue the story using their previous context and their new decision to guide the next scene. Maintain a supportive, accessible, and educational tone throughout, and if any context is unclear, ask follow-up questions to clarify before proceeding.
From the above prototype, I observe the following:
- Although it is not 100% accurate, the narration and choices for CMOA adapted to my context (my occupation, work situation and level of expertise towards my subject matter, etc.)
- Instead of constraining choices of A, B,C,D, I can type the choice freely, and the narration will adapt to the response. I feel it is truly “Choose my Own adventure”
However, currently there is limitation on the following:
- I (user) have to manually input the context. They also misread my context, if I do not input the context accurately.
- The prototype CMOA does not have past data to my learning analytics so the narration won’t change based on how I perform similar assignments in the past.
- Some of the output I’m getting is not accurate. For example, in the CMOA, the narration stated that embedding copyrighted videos is not allowed, but that is not always true from a copyright perspective. AI tools currently experience hallucination, where they generate information that is fabricated but appears authentic (“When AI Gets It Wrong,” n.d.). I am also not sure whether this issue will be fully resolved in the future.
- The CMOA lacks multimedia components such as image, video or audio that makes the narration engaging.
From my experience using the prototype, I believe the future of CMOA can include a personalized CMOA that is combined with learning analytics and virtual reality. Pellas et al. (2021) suggest that future VR learning environments may benefit from the use of data analytics to support personalized immersive learning. In the past, VR systems had limitations in making the experience personalized, but with improvements in AI technology, it is becoming possible to imagine a CMOA where the narration adapts to my past learning analytics data and my identity, and where the visuals change in response to the choices I make and the way I move.
Cartoon-style illustration of a person in VR interacting with a CMOA environment.
Image generated by the author using OpenAI’s ChatGPT (OpenAI, 2025).
However, this also made me question something important. Would EdTech companies begin collecting large amounts of student data in order to enhance the learning experience? And if that happens, will students be able to opt out of data collection without losing access to the full learning experience?
I am curious to know what your experience with the prototype has been, and what you would forecast based on that experience.
References
Reference:
Abbott, D. (2024). Choose Your Own Adventure! An empirical study on gamification of postgraduate learning on research project design. The Journal of Play in Adulthood, 6(1), 18–43. https://doi.org/10.5920/jpa.1475
Cook, E. (2021). Rearing Children of the Market in the “You” Decade: Choose Your Own Adventure Books and the Ascent of Free Choice in 1980s America. Journal of American Studies, 55(2), 418–445. https://doi.org/10.1017/S0021875819001476
Diano, L. M., & Dalanon, J. (2024). Exploring the use of scenario-based choose-your-own-adventure games for restorative dentistry. Journal of Dental Education, 88 Suppl 3, 1959–1961. https://doi.org/10.1002/jdd.13540
Dincelli, E., & Chengalur-Smith, I. (2020). Choose your own training adventure: Designing a gamified SETA artefact for improving information security and privacy through interactive storytelling. European Journal of Information Systems, 29(6), 669–687. https://doi.org/10.1080/0960085X.2020.1797546
Makransky, G., & Mayer, R. E. (2022). Benefits of Taking a Virtual Field Trip in Immersive Virtual Reality: Evidence for the Immersion Principle in Multimedia Learning. Educational Psychology Review, 34(3), 1771–1798. https://doi.org/10.1007/s10648-022-09675-4
OpenAI. (2025). Cartoon-style illustration of a person in VR interacting with a CMOA environment [AI-generated image]. ChatGPT. https://chat.openai.com/
Pelánek, R., Effenberger, T., & Jarušek, P. (2024). Personalized recommendations for learning activities in online environments: A modular rule-based approach. User Modeling and User-Adapted Interaction, 34(4), 1399–1430. https://doi.org/10.1007/s11257-024-09396-z
Pellas, N., Mystakidis, S., & Kazanidis, I. (2021). Immersive Virtual Reality in K-12 and Higher Education: A systematic review of the last decade scientific literature. Virtual Reality, 25(3), 835–861. https://doi.org/10.1007/s10055-020-00489-9
Shemshack, A., & Spector, J. M. (2020). A systematic literature review of personalized learning terms. Smart Learning Environments, 7(1), 33. https://doi.org/10.1186/s40561-020-00140-9
When AI Gets It Wrong: Addressing AI Hallucinations and Bias. (n.d.). MIT Sloan Teaching & Learning Technologies. Retrieved November 13, 2025, from https://mitsloanedtech.mit.edu/ai/basics/addressing-ai-hallucinations-and-bias/
Disclaimer
Although the ideas are my own, I used ChatGPT-5 to improve the grammar and flow of the paragraph. In addition, for the section titled “Prompts I have used to create CustomGPT,” I used ChatGPT-5 to help reformat my prompt so that it would be suitable for creating a CustomGPT.
Hey Rie, neat idea!
Reading your post made me think about how context aware CYOA could feel way more immersive if it leaned even further into mixed reality. The way you described the prototype adapting to your background already touches on the needs for personalization, but I kept imagining what would happen if those branching paths were actually happening around the learner. Instead of just text based choices, AR overlays could guide someone through a scene that shifts depending on their answers. And with VR, learners could step into alternate scenarios that respond to their past performance or habits in a really dynamic way. I keep imagining what this could also look like if you applied this technology to aspects like novel studies or exploring historical perspectives, very cool!
I also share your concern about how much data a system like this would need in order to feel fully adaptive. The corpus would have to be huge! There’s a fine line between personalization and surveillance, and I’m not sure private companies would handle that balance well. I think as technology evolves and we continue to defend and advocate for privacy regulations, we really could see a tool like this become a reality.
Thank you for sharing! I love the idea of using AR as an overlay!
Personalization and surveillance are definitely tricky. I can totally see companies starting off with less surveillance, but later increasing it when monetization gets hard or political landscapes shift, eventually adding more tracking or ads.
Your exploration of context-aware CYOA is fascinating, and the prototype demonstrates the significant potential of generative AI to move beyond rigid, rule-based systems. It’s amazing to see how tools like CustomGPT can facilitate truly open-ended, adaptive choices.
I strongly agree with your emphasis on the storytelling element. In our compliance training materials for employees, we’ve found that learning through narrative is extremely effective. Story not only boosts engagement but also measurably enhances recall and the practical application of the knowledge.
While the integration of VR, learning analytics, and context-aware AI promises an satisfactory personalised learning experience, the roles of educators are also important in the following learning experience, which is to organise a structured debriefing session by the instructor. This step allows learners to synthesise individual experiences and deepen the learning as well. What do you think of my idea? Thanks again for your wonderful presentation.
Hi Makyr,
I agree that the role of educators is important, even when integrating technology such as VR or Learning Analytics. I feel that Learning Analytics sometimes misreads context, and it is important for educators to intervene, analyse, and use the technology to enhance learning for students.
This idea has so much potential. I like the notion of taking an existing format for storytelling and using technology to transform it into something closer to what it describes itself to be. There are so many possibilities for how to use this as an educational tool. It’s an approach to gamification that seems perfectly positioned to foster exploration and learning across many subjects. Really great idea Rie!
Thank you mcober!I also find that AI makes it much easier to build games. One of the biggest challenges in gamification is the time and skill required.
This is an exciting look at how activities could grow with context-focused AI. The prototype was neat and helped show what a more flexible version could look like. Good call on raising concerns about data collection and whether students would still have choice. It’s an important question for any tool that adapts to personal information.
As a Science and Chemistry teacher, I can imagine using a CYOA to guide students through lab safety decisions or exploring different reaction conditions to see how they affect outcomes. A tool that adjusts to their prior knowledge or misconceptions could make those topics easier to learn.
Thank you Nik! I agree that CYOA can be a great tool for checking students’ prior experiences or misconceptions, since the results can be adjusted based on the choices students make.
Hi Rie,
Thank you for sharing your project. I’ve worked with Twine, which is a tool used to create interactive, branching stories where players make choices that shape the narrative, so I immediately connected with this idea of context-aware Choose Your Own Adventure. One of the biggest challenges I’ve experienced is trying to create enough meaningful options for the player without overwhelming the story with endless branches. You want the experience to feel flexible and personal, but you also have to be realistic about how much content can actually be created.
The idea of integrating AI so that the experience adapts to the player’s own context feels really powerful. It could solve the issue of limited choices and make the learning experience feel more relevant and applicable, especially in educational settings where learners come with different backgrounds, experiences, and needs.
That being said, my main concern would be around ethics and data privacy. For a system to be truly “context-aware,” it would need to collect a large amount of personal data, which raises important questions about consent, transparency, and how that data is stored or used. I think finding the balance between deep personalization and protecting student privacy will be crucial if something like this is ever implemented in real classrooms.
Hi chanmi33,
Thank you for sharing your expereince with CYOA.
Twine is a great tool for CYOA especially since it is Free and Open Source. I also agree with the challenge you have experienced. Rule-based CYOA also requires a lot of time to build the branching as well.
I also agree with the concerns around ethics and data privacy. I’m not sure what the optimal solution would be, but I was thinking about either educating students on digital literacy related to CYOA, or keeping surveillance minimal and providing options to opt out.
Hi Rie,
This was an amazing tool that I would love to integrate into my forensics class. If I could develop a crime scene story, students would be able to truly develop their analytical and problem-solving skills through the responsive feedback to their decisions. If it is possible to generate a report about how the students did, that would be even better. If AI is able to provide feedback and suggestions to improve writing, would it not be able to mark a response/ justification of decisions based on a rubric? Is the teacher’s role just to create and interpret what was decided by AI?
I love the idea of generating a report about how students did! I think that will help educators to understand students what concept they are struggling with.
The questions you posted are also very good. I feel that with the introduction of AI, it forces us as educators to rethink the existing assessments we use and review the rubrics we use to evaluate students’ skills.
I feel that even if AI reduces its hallucination problem, our work as educators is still essential, since we are experts in teaching and learning and have more nuanced context about our students.
Thanks Rei. What a unique OER. I never knew you could use chatGPT in such a way. This opens up many possibilities for AI use in context.
As for your question on data collection – this is already happening, and the trajectory suggests it will intensify dramatically. The question isn’t whether data collection will expand, but whether we can design alternative models that provide personalization without surveillance. My ultimate fear is that students will have to choose between sacrificing privacy for educational advantage, or protect privacy at the cost of inferior learning.
Its a difficult question you pose. The better the personalization, the more data required. The more data collected, the greater the privacy invasion. The more privacy protected, the less personalization possible.
Hi Dave,
You’re very spot-on! I also worry about students having to choose to sacrifice privacy in order to learn. I hope in future there will be better solution for this (such as offering more options for opting out of certain privacy settings,improving digital literacy, etc)
Hello Rie,
I had fun using the prototype and liked the flexibility of going beyond the typical ABCD in other “choose your own adventure” type games. I appreciate that you have included a limitations list here, I agree that this would be more engaging for students if images could be used as well. I wonder if this would work better on its own platform rather than using ChatGPT, which likely has limitations for users with free accounts.
Hi Mark,
Thank you! Glad to hear that you liked it.
In future, I am hoping to see CYOA tools such as Twine or H5P to have integration with AI, so that people without free account are able to enjoy context-aware CYOA.
Hi Rie,
I really enjoyed engaging with this tool and think it’s very beneficial for students. It gives them a sense of ownership over their learning, allowing them to explore within the guidelines of a teacher-controlled environment. There are some concerns, such as you, the user, having to input context, which must be done very carefully. However, when done thoughtfully and intentionally, I think this has great potential for student learning and engagement.
Hi Rie,
Thanks so much for making this, I have never used a custom ChatGPT before so this was a great.
Firstly, I thought this was a really fun way to use ChatGPT. The CYOA format made the interaction feel playful, and it was interesting to see how the story shifted based on what you typed rather than fixed multiple-choice options. It’s a nice reminder that AI doesn’t always have to be complicated or reinvent the wheel to impress me.
I thought the prototype was a really fun way to explore digital literacy. There are so many bad ways this information is presented, this felt like dungeons and dragons and I was never quite sure how my choices would play out, I had my expectations but it was actually a little bit exciting versus the standard model (i.e. watch video, > complete multiple choice > receive security certificate)
Thank you kgaudr01!
I think the biggest advantage of being able to type answers freely is that it’s harder to predict what will happen next, which makes the game more interesting.
Hey Rie,
Thanks for getting this working. I love the idea of building your own GPT as a delivery model; I’m sure there are other educators who have messed around with the idea to various success, and it brings the idea of “AI as teacher” even closer to reality.
One piece of feedback, though it is a minor one, is that unless you have an upgraded version of ChatGPT, you get a limited amount of usage out of custom GPTs before you’re cut off for the day. I definitely got the gist of the tool before that happened, just wanted you to know.
There are definitely real issues to consider around what kind of information would be required for AI-assisted technologies to become transformative in terms of personalization. I ran into feedback related to this for my own A3 assignment. On one hand, if no personal information is provided, things are “safer” but less personalized, and I would argue the scope of possibilities is narrowed in this case. As an example; your GPT is simply a repository for digital literacy when taken on its own, but once personal information and context is provided, all of a sudden it turns into something more; a personal coach and dynamic teacher capable of molding its content base to your use cases.
So, to your question about “Will students who opt-out of access lose out on some of the learning possibilities? I would say definitely. Does that mean we should be thinking of ways to make up that gap or build tools from the ground up with safety in mind first? Definitely.
Hi Jakedepo,
Thank you for letting me know about the limitation of ChatGPT. It helps a lot! I hope in future, ChatGPT will ease the limitation for CustomGPT.
I was also thinking about how students can balance privacy and learning experience. One solution could be to provide students with digital-literacy support and give them more agency to choose what to opt in or opt out of based on the pros and cons.
Hey Rie,
I tried to open up your GPT-5 CMOA prototype but am getting an error message saying I don’t have access to that custom GPT. Not sure what the work around is but figured you’d want to know.
Cheers,
Jake
Hi Jakedepo,
Thank you so much for letting me know about it! I just fixed the link – hope this will work
https://chatgpt.com/g/g-69100aa90b68819185f5c64462cf940a-choose-your-own-adenture-digital-literacy
Hi Rie,
Just to let you know the new link is still not working for me.
https://chatgpt.com/g/g-69100aa90b68819185f5c64462cf940a-choose-your-own-adenture-digital-literacy
Hi Mandy, Thank you for letting me know. For some reason GPT reverted back the access to private! Above is the link.