By Sara Feijo and Katherine Ouellette
Generative AI is already changing the education landscape, but can educators ensure a positive impact from this powerful technology?
In a Nov. 29 “Generative AI + Education’’ symposium hosted by MIT Open Learning, leaders in education took more than 250 people on a journey to the future of generative AI in and beyond the classroom. They emphasized how educators must first envision the desired outcomes for children, schools, and society before determining how best to use digital tools to achieve those goals.
Panelists explored how these technologies are transforming the learning experience and teaching practice in K-12, post-secondary education, and workforce upskilling. They also examined the role generative AI should play in supporting effective, engaging, and equitable learning at any age. Key takeaways included:
1. Generative AI has already changed education.
Students are already using generative AI tools like ChatGPT for homework assistance, which alarms educators because they may bypass the assignment’s intended learning objective. For example, essays are often used to teach the mechanics of writing, but learners won’t hone that skill if they’re prompting AI to generate an entire essay for them. Panelists framed this technology as both a potential opportunity and a potential hindrance. If educators reevaluate what they want students to learn, they can revise their curricula to facilitate higher levels of cognitive processing. They can also think about the new opportunities generative AI tools offer to both educators and learners.
However, this isn’t just about technical skills. The Scratch programming language, which is a product of the Lifelong Kindergarten group at the MIT Media Lab, is used by millions of children — and adults — around the world to create and share multimedia projects. But Mitch Resnick, professor and director of the Lifelong Kindergarten group, noted that “we want people to learn about processes and strategies of design that go beyond coding.” Educators should consider how to get students to work creatively, think beyond simple mechanics, and encourage them to ponder deeply about their work.
In a world where information and processes change rapidly and continually, educators and researchers are questioning the value of mere memorization and narrow skills. On the other hand, pedagogies that develop agile learners who are capable of adapting to new and unexpected scenarios are favored by many. Panelists emphasized the importance of fostering opportunities where learners can become creative, collaborative, and curious thinkers. With this in mind, educators could leverage generative AI in their teaching to foster higher-level skills such as critical thinking, analysis, and strategy.
Janet Rankin, director of the MIT Teaching + Learning Lab, said educators should be driven by thinking about what they want their students to do. Once educators know that, they can think about how generative AI fits with those ideas, she said. There have been plenty of disruptive technologies in schools including calculators and the internet. Schools also have a long history of other technologies that had a lot of hype and little impact. Panelists stressed the need to understand which path AI is on.
2. Educators and policymakers must rethink the existing education model.
Many educators and researchers advocate for hands-on constructionist learning, which centers students in the learning process and encourages students to develop their own understanding. However, the instructivist learning model, where teachers deliver instruction to students, is still the dominant education model in many schools. Panelists pointed out that, regardless of the technology at hand, our education system should be moving to more constructionist approaches, where students work on hands-on, project-based learning. With that idea in mind, the question becomes: How can AI support that model?
Modern education balances multiple purposes: instruction, workforce preparation, citizen development, and more. “Historically, technologists have not done a great job of understanding those complex, social, technical systems,” said Justin Reich, director of the MIT Teaching Systems Lab, resulting in new tools developed for the way we wished students learned and schools operated, instead of the ways they actually work. “If you don’t understand these systems you’re building for, then you’re not going to build things that work for those systems,” Reich said.
Pattie Maes, Germeshausen Professor of Media Arts and Sciences at MIT Media Lab, has been thinking about the future and the ways that AI could play a role in learning. When asked about her moonshot, Maes envisioned a context-aware device that is with learners at all times, so its educational assistance would be informed by learners’ experiences. The device could serve as “a mentor, thought-provoker, encouraging you to see things differently and go deeper,” Maes said.
3. Keep equity and access top of mind.
A recurring goal for multiple speakers was to give learners from a broad range of backgrounds control and agency over technology. They expressed concern if these powerful technologies are only developed with limited perspectives, whether it’s a small number of companies in the field or programmers from a narrow demographic. “Who gets included in technology and who does not? What happens when more people participate in tech?” said Randi Williams, research assistant in the MIT Personal Robots group.
Panelists also expressed concern over the increasing disparities that AI technology could bring. If the best AI technologies come with a price tag and take resources to be used effectively, they may privilege well-resourced schools. Panelists stressed the need to think about ways to address these concerns so that AI narrows, rather than widens, existing disparities.
Hal Abelson, professor of computer science and engineering at MIT, argued that generative AI technology should be a tool for everyone, not just highly educated or well-resourced people or those with a technical background. Computational action — a model that seeks to empower children to make a difference in their communities through technology — shows that all children can create tools that improve lives and have meaningful social impacts. For example, high schoolers in Moldova developed a mobile app where people can enter and view clean sources of water on a shared map, a resource that addresses a nationwide problem. Speakers called for the creation of policies to address biases in generative AI and ensure that everyone has access to these powerful technologies.
Generative AI in action
The symposium gave participants a sneak peek into 12 MIT cutting-edge generative AI projects — from K-12 curricula and professional development for teachers about ChatGPT and hidden biases in these tools, to a personalized educational chat tutor for quantum mechanics, to a mobile app that uses AI-assisted observational learning to improve public speaking skills.
The symposium was part of MIT Generative AI Week. This three-day event series explored the latest cutting-edge research, the implications and possibilities of generative AI, and the opportunities and challenges posed by this technology in education, health, climate science, and management.
Visit the MIT News website for the full coverage of MIT Generative AI Week, including an overview of the entire week, a recap of the “Generative AI: Shaping the Future” symposium and a presentation on generative AI-aided art.