Teachers are spending more than 10 hours per week prepping their lesson plans, cutting into time that could be spent with students. A team at the University of Washington wants to help the educators recoup some of those hours and at the same time produce better lessons.
Over the past two years, the group has developed a product called Colleague that uses AI and chatbots to assist K-12 teachers. It’s now available to them for free.
“We’ve really designed this with and for teachers,” said Min Sun, director of Artificial Intelligence for Education in the UW’s College of Education and leader of the Colleague project.
Colleague provides tutorial videos to walk teachers through the process, and chatbots suggest prompts for creating lessons plans. The site offers existing plans that can be edited and customized, or helps educators write their own. Teachers select their grade, subject, and state to align their plans with local standards and curriculum. Colleague also lets teachers create and integrate visual aids using generative AI prompts.
The platform includes a repository of educational resources for teachers to draw from that the UW team has curated and vetted to make sure the material is accurate as well as inclusive for students with disabilities and who are English Language Learners. Absent a tool like this, teachers are left searching and fact-checking online — which is the approach taken by more than 90% of educators, according to a RAND survey.
Colleague can make lesson plans for a variety of subjects, but currently includes extra resources only for math.
Tim Chalberg is a math specialist for middle and high school grades in Tacoma Public Schools, south of Seattle. He was one of the many educators who collaborated with the Colleague team in developing the platform which he said makes “best practices in modern math teaching more accessible.”
An aspect he particularly likes is Colleague’s support for instruction in multiple languages. In the past, he’d spend more than 15 minutes drafting questions in multiple languages for math-related writing assignments. With Colleague, that exercise takes a minute or less and generates helpful new ideas.
“It isn’t just saving time,” Chalberg said. “It’s saving time while also helping me grow in my practice.”
One of Colleague’s unique features is its use of AI to evaluate a lesson plan against 10 parameters, including the relevance of the activities, incorporation of student participation and group activities, inclusivity and overall quality. If the plan falls short, a chat feature suggests specific ways to improve it.
Following OpenAI’s launch of ChatGPT in November 2022, there has been a scramble to develop generative AI tools and the capacity for dialogue between machines and users — including in education.
Seattle Public Schools last month started using AI software from MagicSchool AI, a company selling generative AI products that help teachers generate lesson plans and perform other tasks.
This week, Microsoft announced that it will donate access to its AI cloud infrastructure to allow Khan Academy, an online tutoring site, to offer Khanmigo for Teachers for free to K-12 educators in the U.S. Khanmigo incorporates Khan Academy content and similarly provides support for creating lesson plans, quizzes and related tasks.
Schools and districts have struggled in figuring out how to incorporate AI tools appropriately in the classroom.
After the release of ChatGPT, some feared that AI would be used by students as a means for plagiarism and cheating, leading Seattle Public Schools and others to originally ban ChatGPT from school devices.
Recent research suggests those worries aren’t coming to fruition, and Washington’s Office of Superintendent of Public Instruction issued a high-level roadmap in January to help guide educators, students and families in AI use.
On Thursday, Sen. Maria Cantwell of Washington shared legislation that she’s sponsoring that likewise tackles the question of integrating AI into schools.
Chalberg said he’d recommend Colleague to fellow teachers, but is “waiting for official policies in my district around AI use.”
He expects that district policies will drive adoption, perhaps even more so than a product’s usefulness. “Colleague might have a leg up,” he said, “because they have partnered with local educational agencies well before launching.”
Sun, who is leading Colleague, said the team is building out resources for other subjects, starting with computer science and other STEM subjects. The site will regularly update information on state standards to maintain its relevance.
Sun plans to keep the tool available for free. The UW provided a seed grant to launch the project and the National Science Foundation gave funding for the initial research and development. In the future, the Colleague team will seek support from public and private foundations.
“We’re hoping that more people would be able to really realize that education is a space where AI can really generate a great social value,” Sun said. “Which areas are more important for our society than our kids — our next generation.”
