Listening to the Harvard EdCast conversation with Ying Xu felt a little like someone describing my classroom from the outside. Not the preschool part, obviously, but the bigger reality she names: AI is already part of how young people interact with the world. Whether it is asking SIRI a question, doomscrolling through the continuous YouTube recommendations, or opening ChatGPT/ CoPilot/ Gemini to “just check something,” students are already practicing relationships with AI. School did not start that relationship. We just walked in after it already began.

One thing Xu emphasizes is that learning is deeply social. Children develop language, understanding, and thinking through interactions with other people. Parents, teachers, peers. That point really stuck with me because as a teacher I see the difference immediately when students talk things through with another person. When they explain an idea out loud, challenge a classmate, or ask a follow-up question, the learning becomes visible. AI can simulate parts of that process. It can ask questions and offer hints. Xu even notes that some studies show children comprehend stories better when AI pauses to ask questions during reading.

But she also makes something clear that feels important for educators to hold onto. AI can support learning, but it does not replace the depth of human conversation. In her studies, children still talked more, asked more questions, and drove conversations further when speaking with a person instead of AI. That lines up with what I see with my own students. AI might get them started, but real understanding usually shows up when they have to defend an idea to someone else in the room.

Another part of the conversation that felt very familiar was the discussion about trust. Xu explains that children, like adults, often judge information based on whether a source has been correct before. If AI gave them a good answer once, they tend to trust it again. The problem is that generative AI blends information together in a way that hides where it came from. Unlike a Google search, there are no visible sources. It just presents a confident answer.

From a classroom perspective, that confidence is the real trick. Students see a clean paragraph and assume the thinking behind it must be just as solid. I have watched students bring in AI-generated essays, research questions, or thesis statements that sound polished but fall apart the moment you ask a follow-up question. The issue is not that they are trying to cheat every time. Sometimes they genuinely believe the tool “knows.”

Xu calls the skill needed to navigate this AI literacy, and that idea connects directly to conversations teachers are already having. Students need to know how these tools work, but they also need to know their limits. AI can remix information. It cannot understand context, intent, or truth the way a person can.

What I appreciated most about Xu’s perspective is that she does not frame AI as something schools should simply fear or ban. Instead she suggests something more realistic. AI should support learning, not replace human interaction. That distinction matters. Students might brainstorm with AI, check an idea, or explore a topic. But the deeper learning still happens when they wrestle with ideas themselves and discuss them with others.

There is still a bigger picture here; AI blurs boundaries. When something talks like a person, it becomes easy to forget that it is still just a program. Xu argues that helping children maintain that boundary is important. Students should know they are interacting with a tool, not a relationship.

That idea connects strongly to how I think about AI in school. Technology has always come with waves of excitement and panic. We saw it with television, smartphones, and social media. AI feels like the next version of that same cycle. The difference is that this technology interacts with students in ways that feel conversational and intelligent.

Because of that, simply telling students “don’t use it” is not going to work. They already are. The real question for schools is whether we help them understand it.

For me, that means teaching students to slow down and question what they see. Where might this information come from? Does it match what we already know? What might be missing? Those small pauses in thinking matter. Xu mentions research showing that even simple prompts encouraging students to reflect on credibility can help them evaluate information more critically.

In many ways, the goal is not to eliminate AI from learning but to make students more aware of how they are using it. AI can support curiosity, help students explore ideas, and sometimes provide scaffolding when they are stuck. But it cannot replace the messy process of thinking through a problem, discussing it with others, and arriving at an understanding that actually belongs to the student.

The conversation also reminded me of something teachers know well. Children do not develop in isolation. They exist within families, classrooms, and communities. Technology shapes their experiences, but the people around them shape how they interpret those experiences.

So the real challenge is not raising a generation that can simply use AI. It is raising a generation that can question it, understand it, and keep their human connections at the center of learning. AI might be everywhere now, but the most meaningful learning still happens when students talk, argue, laugh, and think together in a room.

Reading Empowering Learners for the Age of AI felt like two emotions at once: relief and teacher side-eye. Relief because somebody finally said it plainly. Students are already using AI in real life, and schools cannot keep acting like it is a future problem. Side-eye because whenever a framework says “shared responsibility,” I hear, “This is about to become one more thing teachers are expected to master between grades due at midnight and the copier being possessed.”

What I actually appreciate is that this draft does not treat AI literacy like “knowing what ChatGPT is.” It frames AI literacy as knowledge, skills, and attitudes students need to engage with AI, create with it, manage it, and even understand how it is designed, while also thinking about ethics. That matters because it shifts the conversation away from panic and toward agency. Not “ban everything,” not “let AI raise our children,” but “teach kids how to use it without losing their minds or their integrity.”

And yes, the draft basically confirms what every teacher already knows. Students trust confident-sounding information way too easily. If AI says it with enough vibes, it becomes “true.” The framework points out that young people are learning about AI mostly from social media, not educators, and a lot of them struggle to catch when AI invents facts. That tracks. I have seen students treat an AI paragraph like it is a sacred text, then look confused when I ask, “Okay, where did that come from?” Like I just asked them to name their great-grandparents.

The part that hit for me as an ELA teacher is how the draft weaves ethics into everything instead of making it a cute add-on. It calls out bias as baked into human choices about data and design. In other words, AI is not neutral, it is a mirror with a lot of fingerprints on it. That is actually a helpful way to teach it in English class. We already teach students that texts are shaped by perspective, audience, and omission. AI outputs are just texts with invisible authorship and questionable sourcing. So the core skill is the same old literacy question in a new outfit: Who benefits, who gets harmed, what is missing, and why should I trust this?

Where my teacher tiredness kicks in is implementation. I agree with the framework’s vibe, but I also know what happens next. Schools will say, “We support AI literacy,” and then hand teachers a PDF and a prayer. If we want this to be real, students need guided practice inside actual assignments, and teachers need time, training, and clear policies that match real classroom life. Because right now the integrity conversation is exhausting. Not because kids are evil. Because the system is not designed for a world where a tool can produce a decent essay in eight seconds and students think that means the learning happened.

So my takeaway is simple. AI literacy is not a new elective. It is the new air in the building. If we teach students how to question AI, verify claims, explain their process, and protect their own voice, we get learners who can think clearly while AI tries to think for them. If we do not, we get a generation that can produce polished paragraphs and cannot defend a single idea in them. And I do not have the energy to grade that future.