The EDSAFE AI Alliance and Arizona State University’s Mary Lou Fulton College for Teaching and Learning Innovation announced the release of their highly anticipated national paper and benchmark framework, “Future-Proofing Human Flourishing: The Case for a Learning Sciences Benchmark for AI.”
Developed in collaboration with over 100 edtech developers, learning scientists, and civil rights advocates, this paper and benchmark development is a direct response to the rapid, unmanaged proliferation of generative AI in K-12 classrooms. It issues an urgent directive to the education sector to move past surface-level guidelines and establish the immediate public infrastructure needed to evaluate AI tools before they are deployed.
The paper frames Generative AI not as a temporary technological trend to be adopted, but as an arrival technology – a permanent, foundational layer of infrastructure that bypasses traditional gatekeepers. As AI tools shift from backend administrative automation to direct, student-facing AI tutors, the paper warns of an impending cognitive outsourcing crisis. Without a rigorous pedagogical framework, unchecked AI access risks creating a crutch effect that leads to significant declines in independent problem-solving, critical thinking, and long-term executive function.
The newly introduced pedagogical yardstick shifts the evaluation of AI from metrics of simple speed or accuracy to a holistic assessment of how tools align with learning sciences and operate within complex sociotechnical systems. The framework evaluates AI tools across three core dimensions: cognitive foundations, pedagogical design, and measurement accountability.
“Generative AI has fundamentally altered the classroom, but our current implementation has outpaced evaluation,” said Erin Mote, CEO of InnovateEDU. “We face an urgent imperative to ensure this technology serves to augment human potential rather than replace it. AI built for the classroom must do more than remove operational friction; it must actively enhance human flourishing. To do that, we must protect a student’s right to the productive struggle, Socratic thinking, and deep learning that are the birthrights of human development.”
The paper further positions AI literacy and educational data privacy as matters of national security. It argues that a functioning, safe, and evidence-based education system is the bedrock of American democracy and economic leadership. The framework advocates for massive national investments in a robust, public R&D data infrastructure that secures data sovereignty, ensuring that the tools shaping the future workforce are built on pedagogical evidence rather than commercial hype.
“Our work with the learning sciences demonstrates that brain development is inherently nurtured by relationships, environments, and lived experiences,” said Carolyn Trager Kliman of Arizona State University, a key collaborator on the research. “AI companion bots and tutors are reshaping how people connect, confide, and cope, yet they lack the pedagogical judgment and contextual accuracy required to be left unmanaged. By anchoring this benchmark in rigorous, interdisciplinary learning sciences, we are giving researchers, developers, and district leaders the baseline metrics required to protect young minds and ensure technology supports true human development.”
