How can AI boost classroom learning outcomes?

Find out more

Highlights

* Instead of banning AI, leverage it by redesigning assignments, such as coding an adventure game and then using AI to expand its narrative, focusing on student interaction with the tool.
* Shift from simply using AI to critically evaluating its outputs; focus professional development on understanding AI's limitations, biases, and ethical implications within specific subject areas.
* Prioritize "Purpose Over Technology" by defining *why* a subject is taught and what human capabilities are cultivated *before* determining how AI might serve those educational goals.
* Raise expectations for student work: if AI can produce mediocre essays, design assignments that demand depth, critical thought, unique context, and genuine imagination that machines cannot replicate.
* Approach AI integration with an evidence-based mindset, questioning assumptions about cheating or workload reduction, and researching its actual impact on learning processes.
* Implement a "human-in-the-loop" principle, where educators use AI outputs as drafts, applying their wisdom, judgment, and care to refine and improve them, protecting core human domains.

Mentioned

* *Stanford Report*
* Mehran Sahami
* Karin Forssell
* Victor Lee
* Stanford’s Computer Science 106A
* "Infinite Story" assignment
* Stanford’s AI Tinkery
* "Purpose Over Technology" framework
* "Three Ps" of assessment

Transcript

SPEAKER_00 0:00

If this episode makes you think, please let us know in the comments and support us by subscribing and leaving a review. Thank you. Today we are exploring a really insightful piece from the Stanford Report titled Stanford Education Experts Put AI into Perspective. This article brings together the views of several leading experts from Stanford University, Mehran Sahami, Karin Forsell, and Victor Lee, to offer a balanced, evidence-based look at AI's influence on education, weighing both the risks and the immense opportunities while consistently keeping meaningful learning at the forefront. What I found really compelling about this piece is its refusal to get caught up in either the sky's fallen alarmism or the AI will solve everything hype. It grounds the conversation in reality, which, for educators like us, is exactly what we need. The article kicks off with a brilliant example from Stanford's introductory coding course, Computer Science 106A. Instead of banning AI, they have an assignment called Infinite Story, where students code an adventure game and then use something like ChatGPT to keep the narrative going indefinitely. Now for many this might seem counterintuitive. Aren't we supposed to be worried about AI replacing coding? But Meharan Sahami, who's an instructor in that course, has a really pragmatic and I'd say optimistic view. He points out that these tools boost productivity. Industries expect graduates to use them, and let's be honest, students are going to use them anyway, even if we try to ban them. So his approach is to figure out how to teach the fundamentals in a way that truly leverages these tools, rather than just trying to forbid their use. And this to me is classic evolution, not revolution. We're not throwing out the basics, we're evolving how we teach them. This really resonates with the idea that our job isn't to teach students how to outsmart machines, but how to outthink them. It's about building that robust human capacity for judgment and creativity. Think about your own classroom for a moment. Instead of fearing a student might use an AI to draft an essay, what if we designed the assignment so that the AI becomes a part of the process? Perhaps it drafts an essay, but the student's task is to critique it, identify bias, strengthen the argument with their unique perspective and local knowledge, or infuse it with genuine care and imagination that the machine simply can't generate. The real value, as I often say, isn't in what the machine produces, but in how the student responds to it, how they transform it. The piece then dives into a critical question posed by Karin Forsell, who directs Stanford's AI Tinkery. What do we want people to be learning to do? This hits on one of my core philosophies, purpose over technology. It's so easy to get distracted by the shiny new tools, to think, oh here's AI, what can it do for us? But we really need to flip that. We need to start with the why. Why are we teaching this subject? What skills, what understandings, what human capabilities do we want to cultivate? Then we can ask how AI might serve that educational purpose. It's about defining the problem before choosing the tool, every single time. Forsell describes the current moment as a gold rush, with lots of immature products being promoted. And she's right, isn't she? We've all seen those tools that promise the world, but when you actually try them out, they fall short or just automate something that didn't need automating in the first place. This is where AI literacy comes into play. The Stanford experts aren't just talking about using tools, but about understanding their limitations and potential failure modes. Sahami's course, for example, isn't just about coding with AI. It includes discussions about ethics, bias, fairness, and how the training data impacts outputs. This isn't memorizing tool features, it's about developing a collaborative reasoning ability, thinking with AI, not just using it. For a department head planning professional development, this means shifting the focus from how to use Chat GPT to how to critically evaluate AI outputs in my subject area. It's about empowering teachers to understand what the AI knows and perhaps more importantly, what it doesn't know. What assumptions is it making? What biases might be embedded? The article goes on to suggest that AI is an opportunity to raise and stretch our expectations of student work. Forsell puts it wonderfully. I think there's a possibility that we will be revisiting what good writing is with students, so then we will not accept a mediocre essay because that's what this tool can give you. This is powerful. If AI can produce competent but generic outputs, then our assignments must demand more. They must demand depth, care and imagination, making them difficult, if not impossible, to fake. This is where we start to think about the three Ps of assessment, not just the products, but also the process, including how students interacted with AI, and their performance in demonstrating understanding live. We need to design tasks that require genuine cognitive stretch, asking ourselves can AI complete this without the student's unique context, perspective or judgment? If the answer is yes, then we need to redesign the task. Victor Lee, an associate professor at Stanford's Graduate School of Education, really pushes for an evidence-based approach. He notes that many people assume AI leads to more cheating or significantly reduces teacher workload, but he's not so convinced. He urges us all to pause and ask what do we know and where do we have evidence for this? This is such an important call to action, especially in education where narratives can so easily be driven by fear or conviction rather than solid research. His team's early research, for instance, suggests AI changes how people cheat, but not necessarily the overall number. That's a nuanced but vital distinction, and this really touches on the idea of outsourcing your doing, not your thinking. Lee was hopeful that AI would free up teachers from burdensome tasks, allowing them to focus more on students. But he wisely points out that some innovations intended for convenience can unexpectedly lead to more things to manage. He mentions social media and email as examples. This is where we have to be careful about accumulating cognitive debt from over reliance on AI for tasks that really need human strategic thinking. Will AI drafting notes to parents actually save time? Or will it lead to more meetings and more time spent revising those drafts to ensure they convey genuine care and context? This is exactly why something like Stanford's AI Tinkery hosts Tinker Time, where educators get hands-on with tools like AI lesson planners. They're encouraged to ask critical questions. What is the tool good at? What are its limitations? How would you have designed this lesson without it? Did you bring your expertise to bear in a way that improved the output? This is the human in the loop principle in action, where humans remain the ultimate decision makers, and AI outputs are merely drafts requiring critical review. It reminds us that machines can compute, but they cannot wonder, they cannot care, and they certainly cannot replicate the wisdom and judgment of an experienced educator. The article ends by reminding us that while AI is evolving rapidly, the path forward involves staying open to its opportunities while being mindful of potential pitfalls. Victor Lee emphasizes improving research on the nuances of AI literacy, encouraging reasonable relationships with these technologies across a wide population, and teaching critical awareness of bias, privacy, and security. What struck me most was his advice for all of us as educators. I view myself as trying to be responsive to questions that people have right now, and rapid engagement and adoption is pushing AI literacy higher on my priority list. He also advises us to model the behavior we want to see, to not be alarmist, but to be thoughtful. This is fundamentally about change leadership, isn't it? It's about building that know-like trust progression with our colleagues, providing space and time for them to experiment and understand, so that those who might initially seem resistant can become the best drivers of innovation. It's not about the latest platform or advance, as Forsell says, but about clarity regarding what students should learn and designing good experiences to get them there. AI should help us hold the complexity so we have more capacity for creativity, but for connection and for truly impactful teaching. It's about using these tools to deepen learning, not just to automate tasks, ensuring we protect those irreplaceable human domains wonder, care, judgment, relationship, imagination, wisdom, and ethics. That's all for today. Thanks for listening.