AI and Education: Some Key Takeaways

Recently, I attended a discussion organized by Stanford’s Center for Deliberative Democracy on AI for Education (Oct 8, 2023) on AI in education. There were two panels, looking at the teacher and student perspectives, respectively moderated by Stanford students, Caroline Kennedy and Jacob Runben. The panelists included:

• Josh Weiss, Director of Digital Learning at Stanford Graduate School of Education
• Karle Delo, Instructional coach, “coachkarle”
• Matt Scherer, center for democracy and technology
• Wes Chao, computer science teacher at Nueva school

Talking about AI in education from a student perspective, there were several interesting take-aways:

• AI is going to be important in all industries, and we should embrace AI not deny it. AI in education is already here and it is only going to grow. “Students who are not able to use AI are going to be like students who cannot use the internet”
• There were some nice metaphors for how AI could transform education: similar to the introduction of calculators to do math more efficiently, similar to word processors to help write longer reports; similar to computer-aided design to help engineers focus on higher-level things. One panelist observed that having AI was akin to having “infinite interns” and argued that the most important skill around AI was the “taste” or “intuition” in defining problems and using these infinite interns!
• AI cannot replace regular education. In particular, a key aspect of education is to learn critical thinking, and AI should not become a crutch preventing students from learning this important skill. Similarly, an important part of the learning process is the struggle to think through possibilities, and then persevering and coming out with your own learning. It was also observed, that in the history of education and technology and education policy, we have had several prior radical disruptions, but the fundamentals of subjects and the core of direct student/teacher engagement continues to be the same.
• AI in education can be used to reduce some of the time-intensive work around grading tests, etc. Ai can also help in personalizing education, eg. customized feedback. But be careful about what you offload to AI and make sure that it is not core to the educational experience.
• AI can also be used as a study buddy, In many surveys, the relationships and interactions with other students came out as a top consideration in education: AI could be another “personality” that students interact with (friend, family, teacher, coach, AI). One class in Harvard, for example, encourages the use of AI to get different personalities/perspectives on case studies. AI is particularly helpful in brainstorming (“getting started”) and problem solving (“getting unstuck”).
• We need both students and teachers to have a baseline knowledge about the fundamentals of AI, and how to use AI to create something valuable, and the things to look out for (develop the skills to know when to use it and when not to use it). Beyond rules, schools should have clearly published ethical frameworks on how to use AI (“one pager guidelines on ethical use of AI”). We should be careful about bias and discrimination in AI, especially harmful outcomes for students from disadvantaged backgrounds or disabilities.

Talking about AI in education from a teacher’s perspective, many of the same themes reappeared. A few additional take-aways included:

• AI is like any tool that can be used for both good or bad. It is important to have a conversation on both the positive and harmful ways in which AI can be used. It is also important to understand that the technology is still early and we may have both foreseen and unforeseen consequences on societies, cities, environment, etc (the panelist gave the example of how cars revolutionized transportation but also came with effects for a much longer time). Another panelist used the metaphor of electricity to discuss how the same technology can open up diverse possibilities, arguing that there is going to be a lot of trial and error.
• One of the panelists highlighted a nice framework from Educause on AI applications for education for the Four “D”s (I later found the link at EDUCAUSE QuickPoll Results: Adopting and Adapting to Generative AI in Higher Ed Tech | EDUCAUSE Review): Drudgery (helps lighten the load); Design (helps with content); Development (advancing work), and Dreaming (helping think, brainstorm).
• AI can help in customizing content — for example, magicschool.ai has explored approaches where teachers can take a lesson and make it relevant based on specific students. Khnamigo was mentioned as another example where AI was embedded into the teaching experience. One panelist highlighted data that showed that AI can have disproportionately more benefit for students with special needs.
• The panel all agreed that AI would not replace humans in the teaching process. Teachers guide the social process of learning (think of how your favorite teachers inspired you). Teachers handle the messy and unique aspects of teaching (consider the nuances of interacting with different students based on a multitude of cues). One panelist highlighted the skill of metacognition (learning how you learn and learning to teach yourself) to be able to better interact with and benefit from AI.
• Looking ahead, most of the panelists felt optimistic about the technology but with varying levels. Everyone agreed on the benefits in the long run and at the macro-level, but a few panelists suggested a wait-and-watch was better suited to predict the near-term and the micro-level

Overall, it was a great set of panels and I learned a lot listening to the panelists. I was particularly intrigued by the four D framework (enough to Google it and find the original reference!). Overall, the discussions highlighted a complex but promising landscape for AI in education and I am looking forward to seeing how this space develops (and maybe even working on some of these developments! 🙂)