Dave Touretzky is a research professor at Carnegie Mellon University who is focused on cognitive robotics and computer science education. He’s on a quest to overhaul AI education in schools and spearheaded the “AI for K-12” initiative as well as Calypso, an intelligent robot programming framework. He can be found on Twitter @DaveTouretzky, and on his professional website here.
To this point, AI has remained largely absent from K through 12 education. After developing Calypso, a tool that lets kids as young as eight program AI-powered robots, Dave realized there was nowhere for AI learning to fit into the standard curriculum.
Dave set out to change this with the ““AI for K-12” initiative. He’s working to define a set of national guidelines for what students, starting in kindergarten all the way through 12th grade, should know about AI.
The five big ideas are: 1) Perception, that computers perceive the world using sensors; 2) Representation and reasoning, that agents maintain representations of the world and use them for reasoning; 3)Learning, the idea that computers can learn from data; 4) Natural interaction, that intelligent agents require lots of data to interact naturally with humans; and 5) Societal impact, that AI can impact society in both positive and negative ways.
In younger grades where there isn’t room for an AI elective, like K through 6, educators can introduce these concepts in other classes—like science, math, social studies and language.
To educate students on AI, teachers must first be well versed in the technology themselves. But there’s a long way to go before every teacher is equipped to educate their students on AI. The very first questions Dave and his team face from K-12 teachers are: “What is AI? How do you define it?”
Dave isn’t focused on building resources for teachers, but frequently points educators in the direction of helpful tools. For example, educational platforms like Google’s Teachable Machine or TensorFlow Playground are emerging to help both teachers and students understand how AI functions from their own homes. Familiarity with these platforms will help teachers gain confidence in AI concepts and applications.
As AI becomes integrated into grade school curriculum, it’s important to incorporate discussions around the social impact and value of AI. When teachers introduce a topic like self-driving cars, for instance, they should discuss how different groups will be impacted by the innovation. People who can’t currently drive may benefit, while those who make a living off of driving may be harmed.
Blakeley Payne, a research assistant at MIT’s Personal Robot Lab, has already created a curriculum centered around AI and ethics education for middle schoolers. The goal of her course is to provide students with the framework they need to be conscientious developers and users of AI.
Right now, there’s tremendous interest in AI education across the globe. In fact, Dave’s five big ideas have been widely adopted by curriculum developers and his posters have been translated into multiple languages, including Chinese, Korean, Turkish, and Hebrew. Further, in China, there’s a mandate that all K-12 students receive instruction in AI, and educators are actively developing curriculum and guidelines.
Dave emphasized that he doesn’t believe people are questioning the need to educate our younger generation about AI. The question is what’s the best way to do it.
EPISODE 13: DAVE TOURETZKY
Teaching Kids AI in School
Jim Freeze Hi, I’m Jim Freeze, and this is The ConversAItion. A podcast during viewpoints on the impact of artificial intelligence on business and society.
The ConversAItion is presented by Interactions, a conversational AI company that builds intelligent virtual assistants capable of human level of communication and understanding. Today I’m excited to discuss how AI is, and should be, taught in schools for students K through 12. I’m joined by Professor Dave Touretzky, a research professor in the computer science department at Carnegie Mellon. Dave, welcome to The ConversAItion.
Dave Touretzky Thank you, Jim.
Jim Freeze Among other topics like cognitive robotics and computational neuroscience, Dave has done extensive research on computer science education. One of the projects he’s currently focused on is the AI for K through 12 initiative. He’s working to define guidelines for what students, starting in kindergarten all the way through 12th grade, should know about AI. And he’s also working to develop resources for instructors and developers that are focused on a K through 12 audience.
Dave, before we kind of dive into that project specifically, could you start off by just telling us a little bit about your background and in particular, what led to your interest in computer science education?
Dave Touretzky Sure. So I’ve been programming computers since I was 12, which was a long time ago now. I got my undergraduate degree in computer science at Rutgers and did a PhD in computer science at Carnegie Mellon, where I’ve been on the faculty for more than 30 years. And so my whole life has been focused on questions of intelligence and computer science and what it means to be intelligent and how we can create intelligent artifacts. So part of that involves being able to explain these ideas to people. And so my interest in computer science education came from an interest in how we can get people who are not professional computer scientists to think about things like computation and about intelligence, because it’s becoming such a part of our culture, that it’s really important, I think, that people learn how to think about these things.
Jim Freeze I couldn’t agree more. So what inspired specifically the launch of the AI for K through 12 initiative back in 2018? I mean, I suspect that some people would be surprised by the notion that you’d want to introduce the concept of AI starting with kindergartners. So, what were the specific problems that you and your colleagues identified that needed to be addressed?
Dave Touretzky So I’d been developing an AI programming framework called Calypso that lets kids as young as eight program intelligent robots with real artificial intelligence. And I started to wonder, well, how would this fit into a curriculum? And so I looked at the Computer Science Teacher’s Associations national guidelines for teaching computing. So this is the standard, actually, it’s also called the CSTA computing standards for computing education in the United States. And some States have adopted it as an official standard, other States recognize the CSTA standards, but don’t necessarily make an official. But this is what educators look to find out what should we be teaching kids about computing? And in these CSTA standards, there were two sentences about AI and both of those were targeted at the 11th and 12th grade band. There was nothing about AI anywhere else. And I thought this just isn’t right, this can’t be where we want to be.
And so I approached both CSTA, the Computer Science Teachers Association, and AAAI, the Association for the Advancement of Artificial Intelligence and said, we need to do something about this. We need to figure out what kids should know about AI and how we’re going to teach it to them and they agreed. And so that led to this creation of this joint project called the AI for K-12 initiative and then we got NSF to pay for it.
Jim Freeze Oh, fantastic. So as a part of this initiative, I believe there are five key concepts that you kind of developed in the guidelines that you think every child should know about AI. Could you talk a little bit about those five concepts?
Dave Touretzky Sure. So when we started on this project, we decided to model our work after what CSTA had done on the standards. We have a steering committee with originally four people, including Christina Gardner McCune from the University of Florida, she’s the co-chair. We had Fred Martin who at the time was the chair of CSTA, he’s since stepped down. And we have Debra Seehorn who led the standards’ effort for CSTA. And so we wanted to model our work after what they had done it on the computing standards.
And what they did, they have a framework with what they call the five big ideas in computing. And so we decided that to get started, we would come up with the five big ideas in AI and that’s going to be the organizing framework for the guidelines. The guidelines are much more detailed, but the five big ideas in AI fit in a little infographic and we made a little poster and I can just go through them very briefly.
The first big idea is perception, that computers perceive the world using sensors. The second is representation and reasoning, agents maintain representations of the world and use them for reasoning. The third big idea is learning, that computers can learn from data. The fourth big idea is natural interaction, this is a very broad one. It says that intelligent agents require many kinds of knowledge to interact naturally with humans. And when I give talks on this, I often say that humans are the hardest thing for computers to understand because we’re so weird. We’re so wonderful.
Jim Freeze It’s the hardest thing for other humans to understand too.
Dave Touretzky That’s true. That’s true. And then the fifth big idea is societal impact, that AI can impact society in both positive and negative ways. And this has meant to encompass things like the economic impacts of AI technology and also social and ethical issues that are raised by AI technology.
So those are the five big ideas. And what we’re working to do now is to put together detailed guidelines by grade band, explaining what students should know about each of these five big ideas and what they should be able to do with these ideas.
Jim Freeze So when it comes to teaching kids about AI, it strikes me that many teachers probably aren’t that familiar with the technology. Can you talk a little bit about what it takes to equip educators to teach AI to their students and what tools are currently available to help them?
Dave Touretzky Yeah, it’s an interesting problem, because unlike with computing where everyone’s familiar with computers. So when you take a K-12 teacher and you asked them to teach a computing course, you don’t have to explain to them what a hard drive is. They’re all familiar with computers, but they’re not familiar with AI. And so very often the first question we get from K-12 teachers is how do you even define what AI is? What is AI? And then the five big ideas, they’re not familiar with any of those either.
So we tell them that AI is a branch of computer science. It’s the area of computer science that’s concerned with getting computers to do things that when people do them are considered evidence of intelligence. As to what they need in order to teach AI, well professional development is a huge issue. We’re going to need a lot of resources and it’s going to take time to get teachers up to speed with teaching AI.
The good news is that it’s not necessarily expensive because we’re surrounded by examples of AI in our everyday lives. So anybody who has a phone, you can talk to your phone, that’s speech recognition. You can use Amazon for shopping, and that’s doing machine learning to figure out what kinds of things you like and what kinds of products might interest you. You can use Snapchat filters to make your face look like a cat, and that’s doing machine vision and face recognition. So there are a lot of places just in our everyday lives where people can point to AI technology and show students that they’re already living with this. Intelligent assistants, like Alexa or Siri are another good example.
And we’re really interested in finding resources that will run in the browser, so that you don’t even have to install them. You just point your browser at a webpage, and you can use these resources to introduce kids to AI and there are a lot of good resources like that. For example, Google’s Teachable Machine is a fantastic example or for older kids, Google’s TensorFlow Playground is another good example. But there are now many people developing educational resources for AI. So it wouldn’t necessarily be very expensive to introduce AI education, but it is going to take a lot of teacher professional development to get people confident with these concepts.
Jim Freeze Yeah, I would think so. Something else that’s particularly interesting, I think, is how do you envision educators introducing a technical concept like AI to really young students like kindergartners?
Dave Touretzky Well, first of all, when people ask us why would you even try to teach a kindergartner about AI? Our response is kindergartners have spent the last two years talking to Alexa before they show up in kindergarten. So these kids are already quite familiar with AI, but they don’t understand what’s going on inside the box. So we want kids, even from a young age, to have some understanding of what’s going on, to not regard it as magic.
And in the younger grades, K-6 say, there’s not room in the curriculum for an AI elective. So it’s unlikely that there’ll be an AI course that you would take in your very early years in primary school. But what you can do is include AI concepts in your other classes, in your science class, in your math class, even in your social studies class or language class. So, there’s interesting things about AI and language that could be introduced even to very young kids.
Jim Freeze Interesting. You know one of the things you touched on earlier that I think we’ve talked about in a number of episodes of The ConversAItion, which is ethics as it relates to AI. And I know you’ve talked earlier about the importance of teaching ethical design and ethical use of technology, I know that’s a part of the guidelines. How do you recommend introducing those concepts?
Dave Touretzky There’s a very nice curriculum created by Blakeley Payne at MIT that’s used for AI and ethics education in middle school, so that’s an example. One of the things you try to do with teaching ethics to young kids, is to get them to think about the different communities that can be affected by technology and they may have different interests. They may have competing interests. So when you introduce something like a self driving car, you can ask questions like who will benefit and who may be harmed. So for example, people who can’t currently drive may benefit from a self driving car, but people who make their living from driving may be harmed by that technology.
So that’s one of the things we want to teach kids, is just to think about what are the different possible consequences of a technology. That if you focus on just one segment of society, you may miss important outcomes for other groups in that society.
Jim Freeze I think it’s fantastic that ethics are being introduced early into the process. One final question I have for you, how have your efforts been received so far and kind of, where do you think you are in terms of the continuum of creating awareness of these guidelines to getting to ultimately broad acceptance? Where are we in that?
Dave Touretzky There’s tremendous interest in AI education right now, not just in the U.S. but all across the globe. China has mandated all Chinese K-12 students will receive instruction in AI and they’re doing their thing. They’re trying to develop curriculum, develop their own guidelines. We have colleagues in China who we speak with, they’re aware of our work. Russia, Vladimir Putin said back in 2018, that who leads in AI will rule the world, so the Russians have also acknowledged the importance of this.
Our five big ideas have been widely adopted by curriculum developers. We have a poster that explains the five big ideas and that’s been translated into multiple languages, including Chinese, Korean, Turkish, and Hebrew. And so I don’t think anybody’s questioning the need to educate our younger generation about AI. The question is what’s the best way to do it. And what are the concepts that really kids need to understand? And how do we make these concepts understandable to children? Because that hasn’t been a focus in the past, AI has been a graduate level computer science concept, and then it made its way into the undergraduate curriculum. But it’s only in the last couple of years that people started thinking about how you would teach AI concepts to children in middle school and high school. And it’s still a stretch for most people to think about teaching AI in kindergarten. But those kindergartners are really eager to talk to Alexa and to interact with some of these AI tools and so we ought to be meeting their desires.
Jim Freeze I couldn’t agree more. Dave, this has been another fantastic topic that’s manifesting how AI is transforming our society. We very much appreciate you being a guest on The Conversation.
Dave Touretzky Thank you. One last thing, I’d like to point people to our website, it’s aifork12.org, and you can find papers about our work. You can find the five big ideas, including the posters that I mentioned. And we have a resource directory where people can find all of these educational resources, links to things like Teachable Machine and TensorFlow playground, and many other tools that people have been developing. So I encourage people to visit aifork12.org.
Jim Freeze Thank you very much. I appreciate you sharing that.
Dave Touretzky My pleasure.
Jim Freeze Thanks again, Dave.
This episode of The ConversAItion podcast was recorded remotely and produced by Interactions, a Boston-based conversational AI company.
That’s a wrap for this episode of The ConversAItion. I’m your host Jim Freeze, signing off and we’ll see you next time.