Redwood City, CA — New projects will create open, public infrastructure to improve AI-powered literacy and writing tools.
Learning Commons announced $2.8 million in grants to improve AI tools that support educators in providing high-quality literacy and writing instruction. These projects will advance the development of Evaluators that assess the educational quality of AI-generated content and feedback against high-quality datasets and trusted educational rubrics.
“Teachers deserve trustworthy classroom tools that provide high-quality, rigorous content. Tools should deliver content at the right grade level, tailored to each student’s needs, and based on solid learning science to help students grow,” said Sandra Liu Huang, president of Learning Commons. “We’re proud to work with Student Achievement Partners, Quill, and Leanlab Education to create public tools that assess how well AI tools meet these important measures.”
The three grants announced today address enduring classroom challenges: how to share timely and consistent feedback on student writing and how to provide students with literacy content tailored to their development. Many AI-powered edtech tools claim to offer solutions, but they vary in effectiveness, often providing repetitive, generic outputs that aren’t sufficiently challenging and actionable for teachers and students.
“Young writers need lots of practice and specific feedback, which is time-intensive for teachers,” said Peter Gault, executive director and co-founder of Quill, one of the three grantees. “AI-powered tools can support teachers in providing more frequent and detailed feedback, but only if those tools are rigorously evaluated against high-quality standards.”
To raise the quality of feedback that AI tools can offer — and ultimately make them suitable to help teachers give students more frequent and detailed guidance — Quill and Leanlab Education will develop a research protocol and a large, public dataset. The dataset will include teacher feedback on anonymized samples of informative student writing, annotated to highlight effective feedback practices by researchers with expertise in learning science. Developers will then be able to use this dataset to evaluate the quality of current AI tools’ feedback on student writing and train future AI tools to provide high-quality feedback — aligned with research-backed strategies to strengthen student writing.
“Our proximity to schools, students, and educators, paired with a rigorous R&D approach, allows us to ensure that tools of the future are being designed in partnership with school communities. We believe this level of direct engagement with students and educators is a necessary industry standard to build trustworthy and effective AI tools,” said Katie Boody Adorno, founder and CEO of Leanlab Education.
The third grant will support the continued development of text complexity Evaluators. Earlier this year, Learning Commons launched Evaluators focused on literary text for students in third and fourth grades. Developed in collaboration with Student Achievement Partners, the models can read AI-generated text and measure the complexity of the vocabulary and sentence structure. Student Achievement Partners will use this new grant to extend beyond vocabulary and sentence structure to develop a machine-scorable version of the full qualitative text complexity (SCASS) rubrics and use those rubrics to score a dataset of passages for evaluating literary and informational text complexity across all dimensions of their SCASS rubric for grades 3-12.
“To build strong readers in grades 3-12, students need consistent access to texts that are worth reading. Texts that support grade-level comprehension, build knowledge, and invite productive struggle without confusion,” said Joy Delizo-Osborne, president and CEO of Student Achievement Partners. “This work turns the full qualitative text complexity rubric into a transparent, machine-scorable yardstick, so AI tools can be evaluated against research-backed expectations and teachers can trust that the passages and recommendations they receive will actually strengthen comprehension.”
All of the resources created through these investments will be openly available, reflecting our commitment to building public infrastructure for AI in education. These grants build on the early release of Knowledge Graph and Evaluators and our work to translate learning science into the tools, data, and frameworks the field needs to build responsibly.
###
Launched in 2025, Learning Commons builds on the Chan Zuckerberg Initiative’s decade of work advancing learning science and translating research into classroom practice. Through shared, open technological infrastructure built for the public good, Learning Commons aims to better connect the way students learn with the tools they use.