← The Observatory
The AI Co-Author CompanionWeekly Reflective Journal

This Week’s Journal

Before GenAI, There Was Hypertext

This week you explored how hypertext was once promised to democratize learning, and what that history might teach us about GenAI. In your journal entry, reflect on these three questions. You can answer them in any order.

  1. 1
    Which idea from this week changed or complicated how you think about technology and learning? Point to a specific moment, quote or example from the presentation.
  2. 2
    Think of a time you clicked through links, searched online or used a chatbot. Did it feel like real choice, or a “facade of choice”? Who or what shaped the paths that were open to you?
  3. 3
    Second-wave hypertext authors argue that hypertext’s potential lies in co-creation. What would working with GenAI need to look like for you to stay the author of your own learning?

Your writing is saved only in this browser on this device. When you ask the AI for help, your reflection is sent to Claude, an AI made by Anthropic, so it can reply.

Your Words 0 words

The Research That Guides This AI

This companion takes up an idea from this week’s workshop: that technology’s most promising role in writing is co-creation, where the person stays an active author rather than a reader of what the machine produces (Ensslin and Skains). It is adapted from The Undiagnosed: Story Space, a writing app from Dr. Megan Perram’s SSHRC-funded research, and reframed here for weekly student reflection. Open a section to see the studies and what they led to in this tool.

How The Companion Is Designed 5 Choices

Showing Who Wrote What

Draxler and colleagues ran two studies, with 30 and 96 people, on writing with AI. People often felt that sentences the AI wrote did not really belong to them, yet they still put only their own name on the finished piece. The more a person shaped the text themselves, the more it felt like theirs. In a reflective journal, being able to see where the words came from matters for students and for the instructors reading their work.

In This ToolThe AI’s replies appear in coloured bubbles inside the dark companion panel, and your own writing sits in the reflection box, so you can always tell who wrote what. Any words you accept from the AI show in purple in your reflection.

Draxler et al.

The AI Cannot Edit Your Writing

Huang and Saksono designed an AI storytelling assistant and decided on purpose that the AI would help people tell their story but would not write it, so the writing stayed authentic. Chan built a narrative chatbot around a “not-knowing” stance, where the person’s own account stays at the centre and the AI does not hand out answers.

In This ToolThe AI cannot type into your reflection. Its words only go in if you ask it to Suggest The Wording, press Add To My Reflection and then accept the change.

Huang and Saksono; Chan; Draxler et al.

Help With Getting Started

Jennings, a writing scholar with ADHD, finds AI most useful for the parts of writing that are hardest to begin: getting started, organizing and brainstorming. In an eight-week study with 20 students, Nepal and colleagues found that people preferred journaling prompts based on their own lives over generic ones. In a study of 30 people, Park and colleagues found that writing to a chatbot encouraged people to tell their experience as a story.

In This ToolHelp Me Get Started offers one small, specific place to begin, tied to one of this week’s questions and to your own experience, like a single slide or a moment you remember, rather than a big question about the whole topic.

Jennings; Nepal et al.; Park et al.

The AI Does Not Write Your Reflection

People do not feel that AI-written text is really theirs (Draxler et al.). Huang and Saksono, and Chan, both designed their tools so the AI helps but does not write. A reflection is meant to show a student’s own thinking, so a reflection written by the AI would miss the point of the assignment.

In This ToolThe companion will not answer the reflection questions for you. When you are looking for a word, it may offer two or three short options. You choose, and you type it yourself or ask it to suggest the wording for you to accept.

Draxler et al.; Huang and Saksono; Chan

Using Your Own Words

Rettberg shows that AI tends to flatten stories into the most common patterns it has seen. Draxler and colleagues found that the more people shaped a text in their own words, the more they felt it was theirs. Polished, generic phrasing can make every student’s reflection sound the same.

In This ToolThe companion mirrors your own language back to you instead of swapping in polished academic phrasing or its own interpretations.

Rettberg; Draxler et al.

Sources 9 Works Cited
  • Chan, Chitat. “An Exploratory Study of a Narrative Therapy Chatbot with Social Work Students.” Journal of Technology in Human Services, vol. 44, no. 2, 2026, pp. 140–67. doi.org/10.1080/15228835.2025.2559342
  • Draxler, Fiona, et al. “The AI Ghostwriter Effect: When Users Do Not Perceive Ownership of AI-Generated Text But Self-Declare as Authors.” ACM Transactions on Computer-Human Interaction, vol. 31, no. 2, 2024, article 25. doi.org/10.1145/3637875
  • Ensslin, Astrid, and R. Lyle Skains. “Hypertext: Storyspace to Twine.” The Bloomsbury Handbook of Electronic Literature, edited by Joseph Tabbi, Bloomsbury, 2017, pp. 293–307.
  • Huang, Z., and Herman Saksono. “Designing an LLM AI-Powered Digital Storytelling Assistant for Inclusive and Authentic Health Narratives.” GenAICHI: CHI 2025 Workshop on Generative AI and HCI, 2025. Search for this source
  • Jennings, Alex. “Chat(GPT)-ing about the Affordances Generative AI Tools Offer for ADHD Writers.” Xchanges, vol. 19, no. 1, 2025. xchanges.org/chatgpt-ing-affordances-19-1
  • Nepal, Subigya, et al. “MindScape Study: Integrating LLM and Behavioral Sensing for Personalized AI-Driven Journaling Experiences.” Proceedings of the ACM on Interactive, Mobile, Wearable and Ubiquitous Technologies, vol. 8, no. 4, 2024. doi.org/10.1145/3699761
  • Park, SoHyun, et al. “‘I Wrote as If I Were Telling a Story to Someone I Knew’: Designing Chatbot Interactions for Expressive Writing in Mental Health.” Proceedings of the 2021 ACM Designing Interactive Systems Conference, ACM, 2021, pp. 926–41. pure.uos.ac.kr
  • Perram, Megan. Hyperandrogenism and the Marginalization of Pathologized Bodies: Exploring Therapeutic Storytelling in Hypertext Illness Narratives. Routledge, 2026.
  • Rettberg, Jill Walker. “How Generative AI Endangers Cultural Narratives.” Issues in Science and Technology, vol. 40, no. 2, 2024, pp. 77–79. doi.org/10.58875/RQJD7538