The Pulse
Google Gives Teachers AI Tools to Build Interactive Lessons
Google Research has released more than 30 AI-generated learning interactives for middle and high school STEM subjects. The system lets teachers request custom simulations with guided challenges, hints and teacher approval built into the wor
AI.info Team ·
Teachers in an initial U.S. study rated Google’s AI-generated learning interactives an average of 8 out of 10, according to research published September 17 by Google Research. The experiment allows educators to request custom simulations for classroom topics, while a public library already offers more than 30 teacher-reviewed activities for subjects including physics, chemistry, biology and mathematics.
Google’s system uses generative user interfaces, or GenUI, to create educational environments dynamically instead of relying on simulations coded in advance. The company says the work is designed for middle and high school STEM instruction, with teachers defining the topic, reviewing suggested learning objectives and approving the material before it enters the public collection.
The project is described in a Google Research post by Gal Elidan, a research scientist, and Yael Haramaty, a product manager.
Eight out of 10 from 12 U.S. teachers
Google’s initial study involved 12 teachers in the United States. Each teacher requested three custom interactives for specific classroom needs, and the resulting activities received an average quality rating of 8 out of 10.
The researchers say teachers saw the main benefit in adapting activities to a particular curriculum rather than accepting the fixed content of an off-the-shelf simulation. A high school science teacher quoted by Google described the prospect of requesting an activity for any curriculum topic as “amazing,” adding that teachers have traditionally had to accept whatever differentiation a prebuilt simulation provides.
“If I was teaching and I could type this in [for any curriculum topic] and then a simulation would [be generated], that would be amazing... I've never been able to differentiate any of the simulations because it's just, you get what you get.”
High school science teacher, quoted by Google Research
A middle school science teacher said the system was valuable because it allowed educators to “craft and build something that aligns perfectly with instructional goals and learning objectives.” Google does not identify the teachers by name in the research post.
Lessons built around levels, hints and feedback
The generated activities are not presented as open-ended chat sessions. Google structures them as a sequence of challenges that become harder as students progress through a topic. An earth science example begins with temperature-related questions before moving toward more difficult tasks involving rapid warming and storms.
Each interactive can include an introduction that activates prior knowledge, a toolbox containing formulas or theories, several levels of hints, explanations of why an answer works or fails, and worked solutions after a student has tried the problem. The design is intended to support inquiry without immediately giving away the answer.
Google says teachers must approve the learning objectives used to generate an activity. Those objectives can be modified to fit a curriculum before they become the basis for the interactive itself. Examples in the released library include Kepler’s Laws of Planetary Motion, data visualization and projectile motion.
Automated checks are followed by teacher review
Google’s generation process includes self-correcting loops that test more than whether a simulation produces a plausible answer. The company says its checks examine whether the levels cover the stated objectives, whether challenges become progressively harder, whether buttons and controls work, and whether a learner can solve an activity.
One automated evaluation opens a Chrome instance and interacts with the simulation as a user would. The process also tests adversarial behavior, such as moving controls to extreme values. Google says the loops repeat until the generated activity meets its specified criteria, after which teachers review the result.
All activities currently in the public library were generated by AI and vetted by teachers, according to Google. A separate evaluation by STEM teachers in the United Kingdom examined 40 learning interactives. Google says the overall ratings were good or excellent, with physics and chemistry proving the most suitable subjects for simulation-based activities.
Google plans a classroom pilot
The public library is an early release rather than a finished classroom product. Google is working with Google for Education on a pilot in which schools using Google Workspace for Education can request simulations for custom STEM concepts, grade levels and learning goals.
Requested activities will go back to the teacher who ordered them for review. Google says a new interactive can enter the public library only after teacher validation and approval.
The company also plans user-experience research and field studies focused on student engagement and learning gains. Those studies will test a question the early teacher ratings cannot answer: whether dynamically generated simulations improve what students learn, not simply how teachers judge the finished activity.