Teaching the Mind to Reflect: How Generative AI Can Foster Metacognition

This past week I’ve been rewatching old podcasts and TEDs. I came across Dr. Katy O’Brien’s old talk about Metacognition, and one thing really stood out to me. She asked us to imagine being asked to go to the grocery store and get a certain number of things, then she asked the audience to put up their hand if they would write it down as a list. As the number of items increased, so did the hands being put up; I caved on her third list. She revealed to us that our instinct to write the list down did not come from precognition that we would forget, but memory that we had once done something like that before, and forgot! We remembered that we forgot, and that is metacognition.

Metacognition is thinking about thinking – it’s being aware of our own thought process through learned experiences. Metacognitive learners are generally more efficient because they plan, monitor, and evaluate their own learning more effectively. Still related to my previous posts about AI and memory, this time I want to talk about how AI tools are mirroring human metacognitive processes by providing feedback and inciting self-analysis.

The human brain has a feedback loop that involves 3 main elements: executive function (managed by the prefrontal cortex), error detection (in the ACC), and reflection. Each element has its own function in monitoring performance and adjusting strategy. This continuous loop of planning > performance > reflection > correction is seen in multiple cognitive learning theories, like Flavell’s Metacognitive Theory and Zimmerman’s Self-Regulated Learning Theory. Developers are using these theories and the self-regulating feedback loop to design AI systems to support learners by mirroring their own metacognitive learning process.

Let’s look at some real examples. Reflection AI by Hyperwrite is designed to help you create reflective essays, like a diary or a journal. You are prompted to input your thoughts, feelings, and reflections about a certain topic, book, or experience, and I find this very interesting, because I never thought AI would ever pose the question “How does that make you feel?”. If you are an educator interested in applying adaptive learning in your classroom, Smart Sparrow is a cloud-based platform for interactive e-learning content. They also have a great article that you can access here if you’re planning to get started. 

You may think that all this is still experimental and under development, but you’d be surprised to know that there are numerous studies already done to prove the benefits of AI-assisted reflective learning on the human cognitive process. A study by Li et al. (2025) on a cohort of music students demonstrated positive development of metacognitive abilities in the AI-assisted group compared to the control group. When members of the test group encountered technique-related problems during practice, they were asked to combine consultation with an AI tool and advice from the teacher, whereas the control group only had the human teacher to consult. The test group proved more capable of overcoming technical difficulties because the AI tool was able to pull information from their databank and supplemented teacher advice – something that a single human is unable to do. 

With assistance from AI tools, learners can improve focus and attention regulation. They can encode information into memory much better due to active self-evaluation and reflection. They will be more independent and confident, and I personally believe that a combination of cognitive learning and responsible use of AI will bring us much further than we ever thought possible. 

Of course, I will never be tired of reminding everyone that AI is a tool; something to help us perform our tasks more efficiently. There is always danger of over-reliance, letting it do all the work, and instead of strengthening our cognitive function, we end up doing the exact opposite. With data sharing comes privacy concerns, especially how developers use our personal data related to learning analytics, and whether or not it will be sold to third parties and used purely for profit, neglecting the ethics of data use. 

Ultimately, I see us all as explorers in the ocean of technological advances. The current is strong, and it is much too easy to be swept away. Metacognition, the practice of thinking about how we think, is our anchor. Especially in education, we cannot, must not, lose that human touch. The feedback loop of planning > performance > reflection > correction must involve both AI and real human educators with real human experiences. Developers must take reflections from the real human students who use these AI tools, because AI cannot replace human insight and introspection; it was created to enhance it! The future of education lies not just in teaching us how to think, but in helping us understand it, always for the betterment of humanity.