学生自主项目中的GenAI：建议和启示

December 2024 Karen Brennan, Paulina Haduong, Avantika Kolluru, Sally Yao, Jacob WolfHarvard Graduate School of Education Introduction03 Strategies10 Contributors78 “It’s not whether AI is greater than you.It’s whether you plus AI is greater than you.” — Andrew Ho, Professor, HGSE Student-directed projects, projects that are directed by students in content andprocess, can serve as personally-meaningful contexts to develop experience andfluency with a variety of skills, concepts, and disciplines. But it can also be hardfor students to define goals, figure out project scope and tasks, and concretelymanifest their creative visions. “Student-directed” does not mean working completely autonomously. In fact,students working on self-directed projects benefit from a wide variety ofscaffolds and supports. In this moment, generative artificial intelligence (GenAI)is becoming increasingly accessible as a new form of support. When we sayGenAI, we are referring to tools—like OpenAI’s ChatGPT or Harvard’s custom AISandbox—that can generate text, images, or other content, in response to aprompt. Based on our own experiences as teachers and students, we were curious abouthow GenAI could serve as a productive support in student-directed projects. Thisguide, which was generously supported by the Harvard Initiative for Learningand Teaching (HILT), emerged from our exploration of that curiosity. This guide contains advice and inspiration for how GenAI can support self-directedprojects, supporting students as they navigate the uncertainty and opportunity ofwork that isn’t prescribed, work whose outcome isn’t guaranteed or known. Wehope that this guide can support students and faculty in exploring how GenAI canbe used in service of students’ goals and desires. Our process In Fall 2023, we taught the core required course for Learning Design students atthe Harvard Graduate School of Education (HGSE). Students in the coursedeveloped a term-long self-directed project focused on designing a learningexperience. Every student worked on a project—but what they worked on, andhow they worked on it, was up to them. With the new attention to GenAI, we wanted to understand how students in thecourse were using these tools in their project development. We were also curiousabout how faculty colleagues who teach project-based courses were thinkingabout the challenges and opportunities of this particular technological moment.We interviewed 27 HGSE Learning Design students and 7 HGSE faculty members. Before the student interviews, we asked each student to prepare a map of theirproject-development experience, marking major milestones of the journey andindicating what supports and tools were important along the way, including GenAI.This process orientation helped us understand the many different activities thatare involved in a multi-month learning design project. In addition to process questions, we asked students to share their broaderthoughts about student-directed projects and GenAI. The faculty interviews weresimilarly structured, but oriented towards their students’ work rather than theirown use. Of course, these interviews reflect a particular moment in time and a specific set ofstudents. Regarding the particular moment, we hope that this documentation canserve as a record of what felt possible to people at the beginning of expandedaccess to a powerful new technology. Regarding the particular students, weworked specifically with graduate students focused on learning design. But even ifyou don’t work with graduate students or focus on learning design in yourteaching, we hope that ideas about creative, student-directed design projects maynonetheless ignite your pedagogical and design imagination. Advice and inspiration In our conversations with students and faculty, we heard lots of wonderful generaladvice and beautiful specific examples for using GenAI in student-directedprojects. Our aim with this guide is to share both the breadth and depth of what welearned. First, the general advice: Consider the broader impacts.Students and faculty emphasized the importanceof thoughtful use, identifying a number of serious concerns associated with usingGenAI, from its tendency to hallucinate, substantial environmental footprint,accessibility barriers due to cost, potential homogenization of culture, andinherent algorithmic biases. The advice was clear: approach GenAI with intentionand awareness of its limitations and potential harms. Preserve learning and authentic voice.Another recurring theme was the value ofusing GenAI as a support, rather than a replacement, for personal thought, effort,and style. As one student explained, “I think it can be a second brain for you ifyou’re feeling a little stuck. You can ask it to help you get started, but it can’t doeverything for you. You’re still the pilot and it’s just an assistant.” Studentsdescribed the challenges of deciding which tasks to delegate to GenAI and whichto tackle themselves. As