Â鶹ӰԺ

At iSAT, we develop AI-enhanced curriculum units for collaborative learning environments. Rather than treating technology as a standalone tool, our materials integrate AI Partners into the regular classroom experience allowing students to investigate how artificial intelligence works while using it to solve problems together.

Explore an overview of our work below.

AI-Enhanced Curriculum Units

Our instructional materials are grounded in three core principles: meaningful problem-solving through collaborative inquiry, creating psychological safety, and embracing the varied ideas and experiences students bring to the classroom.

Defining Characteristics:

• The Storyline Model: Instruction is organized around a specific anchoring phenomenon or design challenge. This provides "navigational coherence" from the student's perspective, ensuring each lesson builds logically on the questions they generate while investigating the central problem.
• "Under-the-Hood" Exploration: We aim to move beyond the "black box" view of technology. By peering under the hood at data and modeling approaches, students engage in model-building and critique. This helps them develop the practical literacy to know when to trust and when to question AI systems.
• Ethics-Centered Design: Guided by our Responsible Innovation framework, the units integrate ethical inquiry as a foundational element. Students explore issues like algorithmic bias, privacy, and fairness, often culminating in design challenges where they must justify policies based on these concepts.

AI Partners are embedded directly into classroom routines. They are designed to facilitate small-group collaboration, scaffold complex tasks, and assist teachers in orchestrating whole-class discussions. The development of our AI Partners serves as the primary vehicle for the Institute's convergence research.

• CoBi (Community Builder): A relationship-focused partner that supports students and teachers in defining and enacting classroom community agreements. CoBi automatically identifies affirmative examples of collaborative discourse and visualizes them. This allows educators to lead targeted reflections that help students practice and improve their interaction skills over time.
• SPARK (Supportive Partner for Agency, Reflection, and Knowledge-building): A real-time scaffolding agent designed to support small-group sense-making. The SPARK interface includes an interactive workspace where the AI and students summarize ideas together using editable digital sticky notes. It combines typed chat and spoken dialogue, using low-latency speech recognition to provide timely feedback while students debate. It also allows students to provide real-time verbal corrections (e.g.,Ìý"I said 'word', not 'bird'") to help the AI learn and improve its accuracy during the activity.
  Ìý

Research Tools

We use severalÌýannotation schemesÌýin our research to advance our understanding of collaborative interactions between student learners. Supporting tools also consist of theÌýMakeCode Activity Logging PlatformÌýfor researchers and developers to study usage behaviors.

We are committed to releasing our algorithms, software, and curricular materials under open-source licensing, specificallyÌýÌýwhich enables reuse, remixing, and redistribution for non-commercial purposes. Our products pages will be updated with more information and relevant links as they become available. For all other questions, please contact our team atÌýInfo.AI-Institute@Colorado.edu.