Making “think twice” a tangible tool for learning

In a given piece of text, we can find some elements more interesting than others.

Once core aspect of developing creativity and critical thinking is developing your sense of taste—or at the very least forming an awareness of what you’re attention naturally gravitates towards.

From there, the next step is to analyze why you gravitated towards those things… How much of it was intentional? How many of the themes and connections between the ideas did you catch?

One new thing that’s possible with LLMs is assisting in finding connections in language. Alone, students can see what they found interesting by highlighting a piece of text. The AI can pushback and show them examples of things that they missed.

To do that, they can be asked questions like “why did you find what you highlighted interesting?” It can also pull up quotes from the text that could fit the description of what they found interesting, but weren’t highlighted. This reflection can help them build deeper understanding of themselves but also of how they learn.

However, alone the learning can only go so far. There are limits to how much a students confidence and metacognition can develop with expert guidance from an instructor. There are also missed opportunities to learn how to debate, explain, and understand others’ points of view without the social aspect of having peers in the class.

That’s why a system like this would benefit from having an instructor with a special interface. Their interface would give them insights across the whole class and controls to form connections between students. It can allow them to identify common themes or divergent viewpoints that might be worth exploring further. They can even jump in and add custom prompts that would further the students thinking.

Imagine being able to see not just what one student highlighted, but how their selections intersect with others'. For example, if three students highlighted the same sentence but for different reasons, the instructor could prompt a discussion that digs into those differences.

This kind of guided social reflection not only deepens individual understanding but also builds critical skills in debating and considering multiple perspectives.

To visualize how this works in practice, check out this video where I demonstrate the AI-assisted reflection process in a classroom setting. You’ll see how the interface allows for real-time adjustments, giving the instructor tools to make the exercise more dynamic in class and become more responsive to each student's learning journey (similar to what I showed in my Discourse Coordinator UI post).

Moreover, the instructor can use the interface to send tailored prompts to breakout groups, encouraging students to compare their reflections and reasonings. This could lead to a more nuanced understanding of the material, as well as an appreciation for the diverse ways people can interpret the same text.

As students engage with these prompts, they’re not only refining their own thought processes but also learning to articulate and defend their ideas in a group setting. This is where metacognition truly shines—when students are aware of their own thinking and can compare it with others, leading to richer, more informed discussions.