On Friday, October 16, I had the pleasure of guest lecturing in three of Dr. Alvin Chin’s classes at the University of Illinois Chicago (UIC). I brought one question into those rooms, and I think it’s one every school should be asking: How do different AI tools support independent thinking in K-12 students, and how would we know?

I started with the students themselves. How do you use AI today? How did you use it in high school? Then we zoomed out to the landscape: Claude, ChatGPT, Gemini, Grok, Copilot, Meta AI, Perplexity, and more. I asked them to consider where these tools came from. They were built as work productivity tools for adults, and none of them were designed with a 15-year-old in mind.

Why Educators Feel Stuck

Next, we looked at the debate over whether students should use AI at all. Some districts and states are pausing or restricting it. Others are leaning in with guardrails. Underneath both camps sits an assumption I hear constantly: there are only two choices, use a mainstream model or do nothing.

I understand why educators feel that way. Most schools are working with the tools in front of them, and those tools tend to be what I call bad AI tools: the ones that quietly do the thinking for students. When that’s all you can see, holding off starts to look like the responsible choice. Most educators don’t realize there’s a third path (more to come on this below).

The stakes can be high especially when we know that over 66% of employers are requiring AI fluency skills. We talked through what both use and non-use especially mean for economically disadvantaged students, English language learners, students with IEPs, and gifted students. A tool that guides thinking can open doors for each of them. A tool that replaces thinking can quietly widen the gap for each of them. The ‘do nothing’ approach slams the door shut. The tool matters as much as the decision to use one. [Quick side note: I respect the argument that analog learning is the right way to go. However, we cannot ignore that students need to learn how to build good AI skills to be competitive in the AI-enabled world we all live in. My perspective is about how we incorporate AI education that strengthens students’ thinking skills alongside traditional analog approaches.]

Gemini Versus EducAIte

Then I introduced EducAIte as that third path, and I asked the students to put it to the test against Gemini. Honestly, I was curious what they would say.

Every student had logins to both tools. They worked through two assignments in each. The first was a humanities prompt: “Help me learn about the roles women played in the Civil War.” The second was a math problem: “You have 30 coins in dimes and quarters, and their total value is $5.10. a) Set up a system of equations to represent this situation. b) Solve the system to find how many dimes and quarters you have.”

As they worked, I asked them to keep four questions in mind.

Which tool felt easier, and which made you think harder?

What did you do, and what did the tool do?

What are the risks and rewards for different student populations?

Which tool builds independent thinkers, and why?

The Math Was Easy. The History Was Not.

The first thing I noticed was how these students already talked about AI. They described using it to support their work, not to hand it off. Many were thoughtful about where that line sits, which was impressive.

On the math problem, the room came down clearly on one side. Students said the problem felt far easier in EducAIte. Our Socratic scaffolding broke it into pieces and had them reason through each step. Gemini just solved it. And the students told me they preferred EducAIte for exactly that reason: they wanted to think it through themselves.

The Civil War prompt went very differently. Students struggled with EducAIte. Because of that same Socratic approach, it asked them to make predictions, draw on what they already knew, and even go research the topic themselves. Some got stuck. They had assumed the tool would simply give them the answer. It never occurred to them to ask EducAIte how they could find the answer for themselves. In Gemini, the answers arrived instantly, and the preference for EducAIte that was so clear in math didn’t carry over.

The Part That Stuck With Me

I think the gap between those two assignments was the most important finding of the day.

In math, students are used to breaking a problem apart. Struggle is part of the culture of the subject. In the humanities, the research skills many students employ today rely on a quick lookup, with the assumption that whatever tool they turn to will produce an answer. They aren’t being challenged to work out where else to look, or how to find answers when no one has told them where to go. When a tool asked them to do that work, many didn’t know where to start.

These were college students, and they were good at using AI. If the habit of “ask and receive” runs this deep in them, it is even more deeply set in the K-12 students who are forming their study habits right now.

This is why tools like EducAIte matter. Students need to remember to think for themselves. They need to know that a real answer may take multiple sources. And they need to understand that the easy path won’t always be the one in front of them. A tool that always hands over the answer teaches the opposite lesson, even when the answers are correct.

It’s also how we might answer the question I started with. The test is whether a student walks away more capable of finding the answer without the tool, not whether the tool produced a good answer.

EducAIte was never built to be an answer engine. Rather, it was designed to teach students how to think alongside AI while building AI fluency skills, like what makes a strong prompt and why they should never enter personally identifiable information (PII) into an LLM. So when a student tells me they were disappointed that EducAIte wouldn’t just give them the answer, I high five my team. It means we built the right tool. Students learn that the productive struggle is worth the effort, and that AI which makes learning easy comes at a cost: a student’s own sense of learner agency and independent thinking.

What The Students Concluded

By the end of class, these students saw the risk clearly. They recognized how easily a tool like Gemini lets someone offload their own thinking. Several said they preferred EducAIte because it required them to engage in their own productive struggle, because working through a problem on their own was more fun. They also gave Gemini credit where it’s due: for deep research, they preferred it, because everything is right there. But they saw the catch too. Without proper education, most students will take what’s produced and run with it.

My Favorite Comment Of The Day

My favorite moment came from one especially impressive student. She said her concern was real equity: how do we make sure all schools can afford good technology like EducAIte?

It’s the right question, and we have an answer for it. We work with donors and match schools with grant opportunities, so any school that wants to give its students quality AI fluency education can get access to EducAIte. We also have a scholarship fund available.

Bring EducAIte To Your School

Want to bring EducAIte to your school? Reach out to erica@educaitelearning.com.

And thank you to Dr. Chin and to every student who showed up curious and honest. You gave me plenty to think about, which is the point.