The Harvard Graduate School of Education has released a new resource to help educators, parents, students, and policymakers understand how artificial intelligence might help or hinder human development.
The AI Guidance by Age Matrix draws on more than 5,000 pieces of research across 81 developmental areas to show how AI’s effects on learning and human development can differ depending on a learner’s age and what skills they’re developing.
The tool is part of a new initiative the Harvard Graduate School of Education, or HGSE, launched on Nov. 10: Compass Collaborative, which aims to close the gap between what developmental science says about how people learn and the decisions educators, policymakers, and technology developers are making about AI.
Research on the effects of AI tools on student learning is still emerging. Stanford University’s AI Hub for Education’s review of the limited research on the technology’s effects on education shows mixed results and depends on how it is designed and used. A separate audit of meta-analyses on AI’s effects on classroom learning found that the existing research papers are flawed and unreliable as guides for decisionmakers.
Meanwhile, educators are trying to figure out how to incorporate AI into instruction and school operations while minimizing its harmful effects.
“There’s a flood of attempts to understand how to navigate [AI] that’s happening very differently from school to school,” said Jeff Curley, the chief of strategic initiatives in the dean’s office at HGSE and co-director of the Compass Collaborative. “A lot of that is being done without much evidence or research.”
With the Compass AI guidance resource, “we wanted to create something that drew on research but that could be easily understood,” Curley said. “We wanted the resource to be clear and actionable at the point of decision for the educator or parent.”
Though research on AI’s effects on education is muddy, there are “fundamental truths” about how people learn that haven’t changed, said Tessa Forshaw, the faculty chair of the workforce and workplace learning concentration at HGSE and the co-director of Compass Collaborative.
Those truths are: “It’s important that we have proximity to human relationships for learning—so good relationships with more experienced others—but also that we have opportunities to practice the skills that we’re developing,” Forshaw said.
Compass Collaborative used those “fundamental truths” to make “informed hypotheses and perspectives around AI use so we can use [the technology] to support better learning and better development,” she said.
If the AI use case is “moving toward opportunities to practice a skill or toward human relationships, then generally you’re in a really good place,” Forshaw said. If the use case is moving away from those, “then generally that might be something you want to double-think about.”
The resource provides guidelines to use AI in developmentally appropriate ways
The AI Guidance by Age website asks users to plug in their role (such as teacher, parent, researcher, etc.), what developmental domain they want to learn about (such as cognitive foundations, creativity, or relationships), and for which age range (from early childhood to adulthood).
After inputting their choices, users will see a page with a quick take on what the research says about AI’s possible effects on the developmental skills that age range is working on, what Compass Collaborative recommends in terms of how technology can support developing those skills, a few key developmental processes and milestones for that age range, and how AI might help or hinder development at that age range.
For instance, a K-12 educator interested in learning about cognitive foundations for 9- to 11-year-olds will see that in this age range, children are typically still strengthening the cognitive capacities that support learning, such as working memory, sustained attention, and the ability to monitor their own thinking, according to the guidance.
How might AI help cognitive development at that stage? Adaptive platforms might keep the difficulty level in the right range for a student to engage in meaningful practice, according to the guidance. Evaluating AI output might give students practice with critical thinking that passive AI use bypasses, it says.
On the other hand, the guidance warns that AI homework and answer tools might short-circuit the retrieval that builds a student’s memory and leads to learning.
“We really want to make sure that educators are empowered to make a choice that benefits the development of their learners in the moment of instruction,” Forshaw said.
Heather Gauck, an elementary special education teacher in Grand Rapids, Mich., said it looks like a helpful tool.
“There is still so much unknown with AI in education,” Gauck said in an email to Education Week. “This platform provides answers to the questions educators have on how AI may affect our students. This feels like a deep dive into those development stages that are imperative for learning to occur.”
Tim Wells, an assistant principal for Mojave High School in North Las Vegas, said that while “the concept is quite interesting to break down the audience, age, and domain,” the tool still feels pretty basic.
ChatGPT might still be Wells’ preferred resource for figuring out ways to use AI to develop certain skills in students, he said.
“The Harvard site might have more they will develop in the near future, but for now, I would not recommend it,” Wells said in an email to EdWeek.
The AI guidance will be a “living resource,” Curley said. It will be updated as more research on developmental science and on AI’s effects on learning are published.
“We made the decision—which can be hard as an academic institution—to not wait and to try to meet the moment,” Curley said. “Educators and parents are making these decisions in real time today, and we want to share our best understanding, backed by as much research as we have, that shows where there’s evidence behind the use of something or not.”
