Learning by Creating: How Human-Centered AI (2026) Boosts Student Agency

Learning by Creating: How Human-Centered AI (2026) Boosts Student Agency

Learning by creating is reshaping AI in education—not as a tool for automation, but as a catalyst for student agency. As generative AI floods classrooms, many schools risk reducing learning to prompt engineering. In contrast, HAI faculty fellow Hariharan Subramonyam champions a human-centered vision: students don’t just consume AI-generated answers—they build, design, and iterate through creation. This shift transforms learners from passive recipients into active architects of knowledge.

From Automation to Agency: The Pedagogical Shift

Traditional AI applications in education automate grading, generate essays, or summarize lectures—tasks that streamline operations but undermine deeper learning. Subramonyam’s model, grounded in constructivist theory and cognitive science, asserts that true understanding emerges when students create artifacts—code, stories, models, or multimedia projects—using AI as a collaborative scaffold, not a substitute.

Instead of asking AI to write an essay on climate change, students use it to brainstorm perspectives, refine arguments, and simulate policy outcomes. The goal shifts from producing correct answers to developing thoughtful processes. Teachers become facilitators of inquiry, guiding students to critique AI outputs, identify bias, and integrate ethical reasoning into their creations.

How Student Agency Combats AI Dependency

When students create with AI, they develop critical AI literacy—the ability to evaluate, adapt, and ethically apply generative tools. This counters dependency by fostering metacognition: students learn not just what AI says, but why it says it.

• Students design their own AI-assisted storytelling projects, comparing outputs from multiple models
• They build simple chatbots to explain scientific concepts, deepening their own understanding
• Peer review sessions focus on AI-generated bias, not just content accuracy

Case Studies in Generative AI Classrooms (2026)

At Eastside High School, a 9th-grade humanities class used generative AI to co-create historical documentaries. Students selected primary sources, trained a custom model on local oral histories, and narrated final videos—resulting in a 42% increase in engagement and a 31% improvement in analytical writing scores.

In a STEM program in Austin, Texas, students built AI-powered environmental sensors and used generative models to predict local pollution trends. Their project won a national innovation award—and sparked a district-wide AI ethics curriculum.

Building AI Tools, Not Just Using Them

True mastery comes when students move beyond using AI tools to designing them. Simple low-code platforms like Scratch + AI extensions or Teachable Machine enable learners to train models on their own data—turning classrooms into innovation labs.

One middle school in Seattle had students create AI tutors for younger peers. The process required them to map learning gaps, define input-output logic, and debug errors—deepening their grasp of both subject matter and machine behavior.

Barriers and the Path Forward

Despite its promise, learning by creating faces obstacles: educators lack AI literacy training, institutional incentives reward standardized testing, and many platforms prioritize productivity over pedagogy.

Solutions include:

• Professional development in AI pedagogy and prompt design
• Redesigning assessments to value process over product
• Partnering with organizations like Edutopia and UNESCO for ethical AI frameworks

As AI reshapes every sector, education must resist becoming a mere automation corridor. Learning by creating offers a path toward human dignity in the digital age—where technology serves curiosity, not replaces it. The future of learning isn’t about faster answers; it’s about deeper questions, co-designed with AI as a partner. Learning by creating isn’t just a teaching method—it’s a moral imperative for the 21st-century classroom.