With AI tools such as ChatGPT and Claude, we can now use AI as a reflective partner in our self-reflection. AI can serve as a mirror – what I call Human-AI Reflection. In this article, I explain what Human-AI Reflection is and how it can be used for learning.
Related: How can AI be used for journaling?
What are LLMs?
In this article, I will refer to ChatGPT and Claude as LLMs. Technically, ChatGPT and Claude are based on Large Language Models (LLMs). LLMs are AI models trained on large amounts of data, such as social media posts, blog posts, books, and other text documents. Because they have so many examples to draw from, they are remarkably good at contextualizing experiences in the wide variety of examples they have been trained on. This can make it feel like we’re reflecting with a wise person.
What is Human-AI Reflection?
I use the term Human-AI Reflection to refer to when humans and AI reflect together. For example, we can use AI chatbots such as ChatGPT and Claude to reflect on our experiences by asking them questions, sharing our thoughts and feelings, and asking for advice. These interactions with AI can shape our self-reflection.
In digital journaling apps such as Reflectly, Rosebud, and Stoic, AI has already been implemented to guide reflection sessions. AI can pose follow-up questions, can interrupt ruminations, and act as intelligent reflection partners.
In human-computer interaction (HCI) research, some researchers are examining LLMs as an extended mind. In the theory of extended minds, thinking can be distributed between, for example, individual minds, LLMs, and the knowledge bases (e.g., the internet). LLMs can thus be considered an extended mind.
Reflecting with AI is very different from reflecting with simple paper notebooks. Because AI can respond, it can influence our thinking and even change our behavior. This could cause potential harm or unintended effects. For example, LLMs (e.g., ChatGPT and Claude) can embed biases, inequalities, and societal issues derived from their training data. If this gets reflected to the user, it can reinforce injustices.
Related: What is Human-AI Entanglement? A Guide to Fourth Wave HCI and Agential Realism
Reflecting with AI for Personal Development
Due to the capabilities of LLMs, we can use them for personal development. Users are using ChatGPT to study, solve problems at work, reflect on social issues, and ask for advice on their future. This reflects the usefulness of AI tools for our personal development.
I refer to personal development as all intentional actions we take to improve ourselves and to achieve our goals. This requires self-reflection to know what goals we have, our current state, and how to reach our goals. In learning sciences, this is referred to as metacognition and self-regulated learning.
I believe that LLMs can be particularly useful to help us review our lives and to self-regulate through reflections. For example, LLMs can be used for learning purposes.
How Human-AI Reflection Can Be Used for Learning
LLMs can support learning by providing personalized, real-time feedback and offering a safe space to reflect. For example, in education, students can chat with ChatGPT and Claude to learn new terminology and concepts. They can ask questions they might not dare to ask in class. Students can also ask LLMs to research on the internet for them, finding inspiration and examples of similar things that they’re planning to do. They can ask LLMs to give feedback on their work, making it a personal tutor.
Further, students can ask LLMs to help them plan their studies. For example, they can ask LLMs to help them figure out which courses or educational programs they should choose if they desire a certain type of career.
Open Questions
However, even though there are many benefits from reflecting with AI, there are still many aspects we need to learn more about. For example: How much should LLMs “do” for us? If we don’t need to think for ourselves, there’s a risk that we don’t learn the skills we need to become independent thinkers. There’s also a risk of overrelying on AI. And how can we know if LLM-based advice should be trusted? How can we ensure that we’re not given bad advice?
Currently, as a teacher in interaction design at a university, I see that many, if not all, of our students are using LLMs across different aspects of their lives. This raises important questions about how it affects their personal development and whether we, educators, should encourage it. To find a way forward, we must examine and try to understand the possibilities, risks, and potential futures of human-AI reflections.
References and Further Reading
- Extended Minds: The Cognitive Science of Human-AI Collaboration
- Haase, J., & Pokutta, S. (n.d.). Human-AI Co-Creativity: Exploring Synergies Across Levels of Creative Collaboration. https://doi.org/10.48550/arXiv.2411.12527
- Kasneci, E., Sessler, K., Küchemann, S., Bannert, M., Dementieva, D., Fischer, F., Gasser, U., Groh, G., Günnemann, S., Hüllermeier, E., Krusche, S., Kutyniok, G., Michaeli, T., Nerdel, C., Pfeffer, J., Poquet, O., Sailer, M., Schmidt, A., Seidel, T., … Kasneci, G. (2023). ChatGPT for good? On opportunities and challenges of large language models for education. Learning and Individual Differences, 103, 102274. https://doi.org/10.1016/j.lindif.2023.102274
- Reddig, J. M., Arora, A., & MacLellan, C. J. (2025). Generating In-Context, Personalized Feedback for Intelligent Tutors with Large Language Models. International Journal of Artificial Intelligence in Education. https://doi.org/10.1007/s40593-025-00505-6