哈佛教育学院 Compass Collaborative 推出 AI Guidance by Age 矩阵

Much of the research used to inform the AI Guidance by Age matrix predates the advent of generative AI.

Four years after OpenAI debuted ChatGPT, no research-backed consensus exists about how college students should or shouldn’t use generative artificial intelligence tools. 

“This is an unprecedented, experimental wild West moment,” Jeffrey James Curley, chief of strategic initiatives for the dean of the Harvard Graduate School of Education, told Inside Higher Ed. “In higher education, there’s been a slowness to acknowledge that college students are under pressure to often use these AI tools in ways that we know has short-term value but long-term expense.” 

According to one survey, 85 percent of college students say they use large language models to help with brainstorming, research, writing and editing. But whether or not those uses are permissible varies from course to course, because many institutions leave it up to individual faculty members to set the guidelines; while some professors have integrated AI into their curricula, others have banned it entirely. 

Most Popular

• DOD Ends Partnership Between Brown, Naval War College
• Manhattan Targets Mellon for Funding Left-Leaning Ed Orgs
• 3-Year Degrees Grow in Popularity

A lot of those decisions have “centered on technology and not on learner needs, and how they are different at different ages, times and contexts,” said Curley, co-creator of the Compass Collaborative, a research group at HGSE focused on ensuring that “AI is used in ways that support healthy development and learning, as science has long validated,” according to its website.

To advance that mission, the group this week launched the AI Guidance by Age matrix for educators, caregivers and learners. “We’re trying to offer a human development lens on how AI might, in some ways, be used to benefit the learner and accelerate learning, but also in other ways may be quite harmful or have trade-offs that we don’t want to be taking on at the moment,” Curley said.

The matrix allows users to put in their role, choosing from a list of nine, including K–12 educator, higher education educator, workforce educator, student, researcher and policymaker. Then the user picks the developmental domain they want guidance on, such as cognitive foundations, social understanding and interaction, or language and literacy. Lastly, they submit the age range of the students they’re working with. 

Those inputs return a summary of the best uses of AI for that particular circumstance, recommendations for designing assessments, a list of how AI may help or hinder those specific learning goals and additional resources. For instance, the matrix may explain to a higher education educator focused on language and literacy for 18- to 25-year-olds that “AI should be used to widen what a young adult reads, reflect their writing back to them, or help them prepare for a hard text, and leave the composing and the puzzling-out with the young adult.” Then it is up to the educator to figure out how to apply those recommendations in the classroom.

Researchers developed the AI guidance matrix by reviewing more than 5,000 pieces of research on education, psychology and human development. But because there’s still limited independent research on how AI-powered learning tools influence learning outcomes, much of that research predates the advent of generative AI by decades. Nonetheless, the older research has already identified consistent patterns related to how humans learn, which helps researchers assess the likely benefits and drawbacks of new technologies depending on the context. 

“I hope that we start to see more literature in that space, but until we do, we still need something to help us guide the decision-making,” said Tessa Forshaw, a cognitive scientist and researcher at HGSE and co-creator of the Compass Collaborative. “We wanted to put both the emerging and canonical bodies of research together and use that as a starting place to make sense of this moment, rather than just rely on the thin body of knowledge that we have on AI and education.” 

She’s also hopeful that the matrix will push educators to move away from the “broad-brush treatment of the moment” and accept the nuances of AI as a learning tool. 

“Humans don’t love holding multiple things to be true at once, but at this moment the reality is that a tool or technology isn’t inherently good or bad,” she added. “Instead it’s about asking for what, for whom, in what context it’s being used and what it’s replacing.”

Editor's Picks

• Colleges Say OPT Fee Would Cost Tens of Millions
• Democrats’ Texas Takeover Could Reshape Boards, Laws and More
• 5 Takeaways From Democrats’ Historic Night

注意转载授权 · 阅读原文