National study launched to measure how 'agentic' AI affects student agency and learning outcomes

Learning Genie has announced the Marzano & Magaña AI Research Cohort, a national study led by Dr. Sonny Magaña with involvement from Dr. Robert Marzano that will investigate how agentic AI influences student agency and learning outcomes.

By Where Should I Learn Editorial · 2026-09-27

Learning Genie, an AI-powered education platform based in Carlsbad, Calif., announced the launch of the Marzano & Magaña AI Research Cohort, a national study led by Dr. Sonny Magaña that aims to measure how so-called agentic artificial intelligence affects student agency and learning outcomes. The announcement names Dr. Robert Marzano as a collaborator and positions the cohort as an effort to bring systematic research to classroom uses of new AI tools.

The term agentic AI appears frequently in discussions about education technology; it describes systems that do more than respond to commands, instead taking initiative or guiding tasks on behalf of a user. In classrooms, that can mean agents that suggest next steps in a project, prompt reflection, or present differentiated pathways for students. The new cohort will focus expressly on how those kinds of behaviors by AI systems intersect with students’ capacity to make choices, plan, and regulate their own learning.

Student agency — the ability of learners to set goals, monitor progress, and take ownership of learning strategies — is widely regarded by educators as a cornerstone of deeper and sustained learning. Advocates argue that when students are agents in their learning, they develop habits of mind and self-regulation that transfer across subjects. The announced study foregrounds that concept, proposing to evaluate whether AI that acts agentically supports or undermines those capacities.

Learning Genie’s announcement positions the cohort as a national study, but it does not specify key methodological details in the material provided: there is no published timeline, sample size, participating districts or schools, nor a protocol for data collection and analysis. Those details will be important for assessing the study’s generalizability and for interpreting any findings about agentic AI’s relationship to student agency and achievement.

Research into classroom AI needs to grapple with both opportunity and ambiguity. On the opportunity side, agentic systems can potentially scaffold individualized practice, free teachers to focus on higher-order instruction, and surface formative information about reasoning and misconceptions. Proponents say such tools could accelerate personalized learning at scale if they are aligned with sound pedagogical designs.

At the same time, critics and cautious practitioners raise serious concerns. If an AI takes charge of selecting tasks or offering solutions, it may inadvertently reduce opportunities for students to wrestle with problems and make meaning. There are questions about whether student choice becomes meaningful when a system’s suggestions heavily shape available options, and whether frequent AI intervention affects students’ confidence and persistence over time.

Beyond pedagogy, practical concerns remain pressing. Privacy and data governance are central whenever classroom AI systems log student work or model individual behavior. Equity also matters: districts differ in devices, connectivity, and professional development capacity, which can amplify or blunt any benefits an agentic tool might deliver. The announcement does not address these implementation issues directly, underscoring the need for transparent reporting from the cohort as it proceeds.

If the cohort follows standard models for rigorous education research, its findings could inform district decisions, teacher professional learning, and vendor design. Research that distinguishes between short-term gains in performance and longer-term development of agency would be particularly valuable. Likewise, studies that surface how teacher practice mediates AI effects — for example, which teacher scaffolds preserve student choice while leveraging agentic features — would offer practical guidance to schools piloting these tools.

For now, the Marzano & Magaña AI Research Cohort is a development to watch: it attaches two recognizable names in education research to a question that is moving rapidly from theory into practice. Learning Genie’s announcement opens the door to systematic study, but reporters and practitioners will need access to the cohort’s study design, measures, and eventual results before drawing firm conclusions about agentic AI’s classroom impact.

The cohort’s origin in Carlsbad, Calif., and its framing as a national study suggest the organizers intend broad engagement, but the value of the effort will depend on methodological transparency and peer review. As districts and vendors race to deploy AI-enabled tools, measured evidence about how those tools influence student agency and learning outcomes is precisely the kind of public-interest research education leaders have been calling for. Follow-up reporting should seek the cohort’s protocols, participating partners, and timelines to assess whether the study can deliver that evidence.