I recently asked my 13-year-old daughter, a Seattle Public Schools seventh grader, how much she was currently learning about and/or using artificial intelligence. “Not at all?” she sort of asked me back.
For a kid with an iPhone who interacts with Amazon’s Alexa on a daily basis, Kate is using AI more than she realizes. But aside from a STEM class she took as an elective in sixth grade, where she learned how to code a simple game, she was not getting any formal AI or computer science education this year at her middle school.
So I checked in with Karim Meghji, the new president and CEO at Code.org, the Seattle-based computer science education nonprofit, about whether I should be concerned that Kate is going to get left behind and never command a $500,000 salary at OpenAI.
Meghji is well positioned to know the answer. The tech vet spent 10 years at RealNetworks and is a former CTO at Seattle digital remittance company Remitly. He joined Code.org in 2022 to serve as chief product officer, leading a shift toward an AI-centered strategy at the organization, which was launched in 2013 by brothers Hadi and Ali Partovi with a goal of expanding computer science education to K-12 students.
Today, Code.org says its AI curriculum has already helped more than 6 million students learn foundational concepts around the technology, and more than 25 million students have completed activities in its “Hour of AI” campaign.
But while Meghji did briefly recommend some learning Kate could do through the Code.org curriculum, his goal was not promotional, and our conversation felt more like two parents, or a parent and a teacher, discussing the fundamentals of understanding a technology that will surely continue to shape the society Kate lives and works in.
A “glass box” approach: Meghji believes that middle school serves as the ideal transition point from basic AI literacy to true fluency, moving beyond introductory games to a “tinkering age” where students can finally look under the hood. Just as students dissect frogs to understand biology, Meghji believes this is the stage where they should begin “dissecting” AI models to understand the data and logic driving the technology.
“AI is this black box for most people today in the world. You put a prompt in, you get something back out,” Meghji said. “Our perspective is it needs to be a glass box, and we need to give them a screwdriver and a hammer and let them kind of get in there and unpack this thing.”
Students should learn about different kinds of data and how it all works, including the different ways information is input into AI models through prompts and context, and the how and why behind AI’s different outcomes.
Beyond the technical aspects: People who are older and allegedly smarter than Kate are fooled every day by content created by AI tools. That’s not going to slow down.
And Meghji said students today increasingly need to work in digital spaces where they understand such such things as how humans factors relate to design, and the ethics around AI.
“These are two very important components that are not technical, but about the application of technology that is really critical as more of us become builders and creators,” he said.
Students learning about technology also need to learn intangible skills that are important in almost any discipline. How do you work together? How do you communicate? How do you collaborate? How do you reflect, so that you can make something better?
What if she’s not going to be a software engineer? Kate shows more interest these days in ceramics than computer science, and I have a shelf of misshapen coffee mugs to prove it. But I’m smart enough to know — and have interviewed enough startup founders who agree — that AI is rapidly impacting all sorts of occupations and businesses.
“AI fluency and computer science foundations are so critical to just about every work experience you have,” Meghji said. “It doesn’t matter if you’re a software engineer, a biologist, a doctor, an architect, you are interacting and working with systems and tools, likely for a good portion of your day. And you have an ability to actually not only use those tools in the work you’re doing, but to make them better and make your work better if you understand what’s going on.”
So falling behind doesn’t just mean Kate isn’t learning how to build her own AI large language model, it means she might not understand how to use one to make herself a better doctor or welder.
Code.org aims to move students beyond “low literacy” interactions, which Meghji describes as simple, one-off prompts that treat AI like an “evolution of search.” Instead, he wants students to engage in deep, multi-step dialogues where they challenge the tool and inspect its “chain of thought” reasoning. By learning to guide and refine these autonomous systems, students can move from merely using technology to effectively collaborating with it.
Final tips for where to start with kids and AI: There are many things about AI that concern and confuse me, so pushing my kid toward adopting the skills to use the tech has admittedly been a slow roll. Meghji offered up a few ways to get started:
- Experiment together in low-stakes ways: Sit down as a family and explore AI tools for text, images, and video as a team. The goal is to find a child’s specific passion and layer on a “parental construct” of guidance, ensuring they learn to use these tools responsibly rather than just consuming them in isolation.
- Advocate for computer science in the classroom: Meghji encourages parents and students to ask for specific curricula, like Code.org’s “Computer Science Discoveries,” which teaches middle schoolers to build games and websites while working directly with AI models.
- Embrace the “tinkering age”: He recommends using platforms like Scratch or Code.org to keep coding skills fresh through play. By “experimenting with the technology of building” — whether through block-based coding or generating apps with AI tools — students develop the builder mindset necessary for any future career, “whether she’s a welder or a cancer researcher.”
