Authors: Alexis Andres, John Whitmer, Melanie Kurimchak
https://doi.org/10.35542/osf.io/ngbkv_v1
Abstract
This study was conducted to ensure that emergent efforts to build an AI infrastructure for education technology are aligned with real-world needs. We conducted 15 interviews with 22 key stakeholders from Digital Learning Platform providers and R&D teams in the Learning Engineering Virtual Institute. We also conducted a literature review of 26 reports and documents related to AI and education. The research reveals a complex landscape where market pressure drives rapid AI adoption, while concerns about the quality of AI-powered solutions create significant barriers to responsible wide-scale adoption. The findings highlight significant gaps in current AI evaluation frameworks for education. Our findings are succinctly summarized in one stakeholders quote: “it’s easy to get something from GenAI, but it’s hard to get something that you would stand behind professionally.” It bears mentioning that this individual had worked directly with researchers in Frontier Labs and had a substantial R&D team with AI expertise. Critical infrastructure needs include evaluation and quality assurance tools, privacy and security solutions, contextualization frameworks, and specialized applications like classroom-optimized Automated Speech Recognition. The report identifies potential complementary roles for philanthropic funders and Frontier Model1 providers, where philanthropy can support public goods like evaluation standards, research-backed implementation practices, careful data annotation processes, and equity-focused initiatives, while Frontier model providers suggest best-in-class approaches at using LLMs, identify new capabilities, and propose methods to automate these approaches. The research confirms that contextualized, research-backed applications of AI are difficult to create and scale, validating the core premise that education-specific AI infrastructure is essential for meaningful educational outcomes. These
findings are intended to provide market guidance to future development of AI infrastructure tailored to meet the evolving demands of the EdTech landscape and contribute to efforts to increase access to digital public goods that improve AI implementation quality and educational equity.
Suggested Citation: Andres, A., Whitmer, J., & Kurimchak, M. (2026, March 2). Not Ready Yet AI Infrastructure EdTech Market Research. Retrieved from osf.io/preprints/edarxiv/ngbkv_v1