#244 – TeachAI, with Dr. Kristen DiCerbo

Unpacking Education December 13, 2023 46 min

In this episode, Dr. Kristen DiCerbo, Chief Learning Officer at Khan Academy, joins us to talk about some exciting new artificial intelligence (AI) resources available to schools and educators. We explore the AI Guidance For Schools Toolkit, which was developed by TeachAI. Dr. DiCerbo also shares some insights about Khanmigo, the artificial intelligence tutor and teaching assistant being developed by Khan Academy. Her experience and work on these projects helps to shed light on the rapidly developing world of artificial intelligence and provides insights into pathways forward for schools and educators.

Read a transcript of this episode.

Dr. Kristen DiCerbo
Chief Learning Officer at Khan Academy
Paul Beckermann
PreK–12 Digital Learning Specialist
Rena Clark
STEM Facilitator and Digital Learning Specialist
Dr. Winston Benjamin
Social Studies and English Language Arts Facilitator

It is in a spirit of humility that we offer this toolkit. My sincere hope is that teachers feel guided and supported by their leaders as we all adapt to the changes AI brings to education.

Pat Yongpradit, Chief Academic Officer for Code.org and Lead of TeachAI

Resources

The following resources are available from AVID and on AVID Open Access to explore related topics in more depth:

Unpacking the AI Toolkit

The AI Guidance For Schools Toolkit, a free resource from TeachAI, is the result of a collaborative effort by leading educational organizations, including Code.org, ETS, ISTE, Khan Academy, and the World Economic Forum. These groups pooled their resources to develop a collection of materials to help schools and educators navigate the rapidly changing landscape of artificial intelligence and its impact on education.

This episode includes insights into artificial intelligence in schools as well as how this toolkit can be used to help school systems reduce the workload in adapting to the changes brought about by AI. The following are a few highlights:

  • About Our Guest: Dr. Kristen DiCerbo is the Chief Learning Officer at Khan Academy. She explains her role at Khan Academy, saying, “I lead the content, product, design, and community support teams. Basically, the experience that you have when you go to Khan Academy is mostly designed and created by folks that are on my teams.” Over the past 12 months, her teams have been working to develop Khanmigo, an AI tutor and teaching assistant. In addition, she says, “I’ve spent almost 20 years now thinking about designing digital learning experiences—and how we build education research and what we know about how people learn into those experiences.”
  • TeachAI: TeachAI is a nonprofit effort to help schools keep up with the rapidly changing world of artificial intelligence. Code.org, ETS, ISTE, Khan Academy, and the World Economic Forum collaborated to form the steering committee and have since been joined by many other entities to generate content and guidance that is freely available on their website at TeachAI.org. Dr. DiCerbo says, “We knew there was a lot of learning to be done, and . . . we were thinking about what teachers need to know to be able to incorporate any kind of artificial-intelligence-based tool into their classroom.”
  • The Toolkit: The AI Guidance For Schools Toolkit is free to use and was released on November 7, 2023. Realizing that schools were struggling to keep up with the rapid changes brought on by generative AI, Dr. DiCerbo says that TeachAI is “hoping that we can, together, as a group of organizations, be able to provide some help and tools for teachers and school administrators to be able to get their arms around what AI is.” The toolkit provides guidance, example language, and sample implementation plans that schools can use to help them on this journey.
  • Benefits of AI: Dr. DiCerbo says, “I think it’s helpful to think about: What are the learning problems that we know about, and can this technology help us solve any of those?” This problem-based approach gives purpose and direction to the integration of AI into school programs. Two areas of need that AI has the potential to address are immediate student feedback and student support. For teachers, it has the potential of saving them time with planning and preparing for lessons.
  • Favorite Part of Khanmigo: Dr. DiCerbo’s favorite feature of Khanmigo is an activity called “Craft a Story,” where the student and AI trade off writing sentences to construct a story. She finds this collaborative writing activity to be both fun and engaging.
  • Potential Risks: It’s important for educators to understand the limitations of large language model (LLM) AI programs. As Dr. DiCerbo points out, they can be wrong and make up, or “hallucinate,” answers. She reminds us, “Users absolutely need to go in being aware. They should not assume that the answer they get from these models is the absolute truth.” She also points out that Khan Academy has worked hard to mitigate these shortcomings within the Khanmigo workspace.
  • Seven Principles of AI in Education: The toolkit outlines seven principles to consider in developing AI policies and practices within a school system. Each principle includes questions to consider, a description, and real-world examples.
  • Try It: Dr. DiCerbo encourages teachers new to generative AI to try it out. Pick a popular AI chatbot, like ChatGPT or Google Bard, and experiment with it to get a sense for how it works.
  • AI Detectors: Dr. DiCerbo points out that AI plagiarism detectors really don’t work, and perhaps most concerning, they can lead to false positives that accuse innocent students of cheating. She says, “They tend to give false positives to folks that are English language learners and folks . . . that write in more . . . structured ways and have less flowery language; they’re actually going to trigger that and give a false positive.”
  • Modifying Assignments: Rather than using AI detectors, educators should consider how assignments can be reworked to either intentionally use AI—while citing how it was used—or tie tasks into specific classroom experiences that an AI chatbot cannot know about. Documenting how they are using AI “really encourages metacognitive thinking,” which can be an added benefit of this approach.
  • Khanmigo and Academic Integrity: AI tools that are designed specifically for education often contain guardrails that are not present in the more general AI chatbots. For instance, Khanmigo has written safeguards into their prompts through the process of prompt engineering. These commands direct the AI system to help students complete a task while not providing them with the specific answers. In many ways, Khanmigo has been trained to act like a tutor. Transcripts and chat summaries are also integrated into the platform to help teachers keep track of student activity.
  • Three Stages of Implementation: The toolkit offers a three-stage approach to implementing AI into schools at a systems level. The stages include addressing such questions as, “What are the areas that we think this technology can help us with—and maybe be a little more transformative and help us scale the use of this technology throughout the school—so that we can get the benefits of it and reduce the risks of it?”
  • Computer Science (CS): CS education is quickly becoming a foundational knowledge area for students. Dr. DiCerbo points out, “People need to have a basic understanding of how technology works. We are so connected to technology these days that if we don’t have some understanding of how these machines work, we could be headed for that dystopian future we’ve all heard about.”
  • Moving Forward: AI development is moving incredibly fast, and it’s hard to predict the future. In that light, Dr. DiCerbo suggests one guiding question that will continue to be relevant regardless of where the future leads. She says we should ask, “When is it most useful, and how do we integrate it into our workflows when it makes sense to do so?”

Guiding Questions

If you are listening to the podcast with your instructional team or would like to explore this topic more deeply, here are guiding questions to prompt your reflection:

  • What do you know about generative AI?
  • What questions do you have about integrating AI into schools and classrooms?
  • What are potential benefits of AI in education?
  • What are potential risks of AI in education?
  • What aspects of the AI Guidance For Schools Toolkit sound most helpful to you?
  • What are your impressions of Khanmigo?
  • What are your next steps in addressing AI integration in your school?

Extend Your Learning

Did you find this resource useful?

Your rating helps us continue providing useful content in relevant subject areas.