AWS Public Sector Blog
Streamlining educational assessment tools with generative AI on AWS
Curriculum Associates is an education technology (EdTech) company using data-driven technology to bring personalized, evidence-based learning to K12 education institutions. Their personalized assessment and instruction products serve more than 13 million students and over 1 million educators around the world.
Given the depth of personalization presented to students and teachers in these offerings, Curriculum Associates builds their software with a variety of data taxonomies informed by dynamic K12 educational content, state and local education requirements, and more. When Curriculum Associates wanted to streamline the process of aligning educational content to different taxonomies—a time-consuming effort for their staff—they turned to Amazon Web Services (AWS).
Drawing on the AWS team’s considerable AI expertise and using generative AI resources from AWS, Curriculum Associates transformed this personnel-intensive alignment process, freeing up their staff to focus on efforts more directly related to supporting student growth and serving teachers.
Meeting the challenge of dynamic state assessment requirements
Curriculum Associates’ core offerings include the i-Ready Assessment and Personalized Instruction programs. Over 13 million students use i-Ready today—one third of K-8 students in the US. i-Ready Assessment is a coherent assessment system that helps teachers meet their students where they are in their learning journey. Educators administer i-Ready assessments periodically throughout the school year, assessing how a student’s performance aligns to the state’s education standards on a given topic. Then, by incorporating i-Ready Personalized Instruction, students receive a personalized curriculum based on their assessment results to support their learning growth exactly where they need it most. Teachers get actionable data and a complete picture of student performance relating to their grade level and national norms, plus a snapshot of overall class performance and progress in relation to state standards.
For Curriculum Associates, continuously updating and customizing i-Ready requires considerable time and energy. The team must align its instructional recommendations to dynamic education standards, which differ from state to state and can change over time. Historically, subject matter experts (SMEs) reviewed the company’s data taxonomies of skills and instructional content and manually aligned them to each state’s standards.
“The taxonomy alignment is a huge effort,” says Dan Mix, associate vice president of assessment at Curriculum Associates. “Developing that across 50 sets of state standards for both math and reading involves hundreds of thousands of connections. It is labor intensive for us.”
Inspiring efficient solutions with an AWS incubator
After Curriculum Associates engaged in a company-wide effort to update their internal assessment taxonomy for use across their products, they needed to realign their content to state standards. To address this resource-heavy undertaking, the Curriculum Associates team considered how to make the process more efficient.
In conversations with its AWS business development team, the Curriculum Associates development team was introduced to an incubator opportunity to work on a generative AI proof of concept (POC). The POC was chiefly tasked with aligning multiple taxonomies, such as state standards, and then integrating these with the company’s content taxonomies. Then, the POC needed to produce aides, like rationales and coverage analyses, for employees editing the taxonomies and content, using AI alongside human expertise.
The Curriculum Associates team developed its generative AI POC in Amazon Bedrock, a fully managed service that offers a choice of high-performing foundation models (FMs), to improve the overall taxonomy alignment. The POC also uses Amazon OpenSearch Service to facilitate search functionality across the taxonomies and the company’s content.
Streamlining process improvements with generative AI
While still in the POC stage, Curriculum Associates’ generative AI solution has already offered multiple improvements. Chiefly, the POC drastically reduces the time SMEs spend mapping between the different taxonomies and content. “Previously, a SME would have to manually go line by line and try to align these things,” says Dean Serrentino, senior director of AI at Curriculum Associates. “[The POC] speeds up the process and lets them more quickly dig into the details… We’re working 3–5 times faster now.”
“One of the capabilities gen AI added to the manual alignment process ties into new state standards,” explains Jeff Stillson, principal data scientist at Curriculum Associates. “It makes evaluation much easier for us now. When a new taxonomy comes out from a state, we can quickly determine how much of it is covered by the items we have now. We then have a good idea of what areas we need to generate new items to fully cover that new state standard, and that makes that process much more efficient.”
The generative AI POC also opens up new opportunities to serve educators better and faster. “If we are working more deeply with a particular state, we can be more proactive and quickly get a sense of how different their skills taxonomy is to something we already support,” says Serrentino.
Plus, by handing off smaller details to the generative AI POC tool, Curriculum Associates SMEs can tackle higher-level challenges. “They don’t have to keep going back and forth between taxonomies,” says Mix. “It allows them the brain space to make better decisions. The quality of the alignment is improved, not only because you’re getting verification from the AI but also because it’s creating this mental efficiency. Many of the details have been taken out so subject matter experts can focus on higher-level decisions.”
Now, Curriculum Associates staff and editors can redirect their resources and talents into developing more valuable tools and supports to impact teaching and learning in classrooms nationwide.
Building on AWS familiarity and security
For Curriculum Associates, long-term familiarity and established relationships with AWS made it a simple decision to reach out for support on the taxonomy alignment project. “We already had a lot of internal expertise within the engineering and DevOps group on working with AWS tools,” says Serrentino.
Curriculum Associates also valued the security that AWS provides because it was critical to keep data in an environment with stringent security policies.
“We needed to make sure we were running in a private model instance we trusted,” says Stillson. “What we’re working on is very proprietary data for Curriculum Associates, all of the item data. We didn’t want it going external to a controlled environment outside of our sphere.” AWS is backed by a deep set of 300 security services and features trusted by millions of customers, including the most security-sensitive organizations like government, healthcare, and financial services.
Building a foundation for future improvements with generative AI
With their POC yielding results for the team, Curriculum Associates is deciding how to further develop its taxonomy alignment process. The next step will be integrating these tools across the company so that content editors have access to these new capabilities.
“Working with AWS really catalyzed this project for us, allowing us to quickly get off the ground something that would have taken us a lot longer by ourselves,” says Serrentino. “It allowed us to jump in and start building prototypes quickly and allowed us to show the rest of our team that this approach is going to work.”
Looking ahead, Curriculum Associates is focused on advancing continuous improvement across the company to create more opportunities to serve educators and support meaningful learning gains for all students.
Learn more about how AWS supports EdTechs to start, expand, and optimize their business as they deliver on their mission for education.