A practical approach to keeping human judgment at the center of classroom AI
I did not start ThinkTorch AI because I wanted more AI in classrooms. I started it because AI was already there, and I wanted a better way to decide what we should actually do with it.
As an English teacher, I can ask generative AI to brainstorm an argument, organize an essay, revise sentences, suggest feedback, or produce an entire draft. That makes the technology useful. It also makes it remarkably easy to remove the exact thinking I am trying to teach.
That tension led me to develop ThinkTorch AI, a platform for helping educators use AI without giving up the thinking that matters. At the center of its classroom approach is the AI Learning Innovation & Growth Network (ALIGN), an instructional framework for deciding when AI belongs in learning and what role it should play. The question I keep coming back to is not, “How can I add AI to this lesson?” It is, “What thinking needs to remain human, and where can AI help without taking that thinking over?”
Start With the Learning, Not the Tool
AI integration can go wrong before students ever open a chatbot. A new tool appears, we see what it can do, and then we start looking for places to use it. The technology becomes the starting point instead of the learning.
ALIGN reverses that order. First, decide what students are supposed to know or be able to do. Then identify the thinking that develops that skill. Only after that should we decide whether AI belongs anywhere in the process.
If students are learning to develop a claim, having AI generate the claim may make the assignment easier, but it also removes much of the work the assignment was designed to teach. If students develop the claim themselves and later use AI to challenge their reasoning, the same technology serves a very different purpose.
Sometimes AI should play a meaningful role. Sometimes it should have a narrow role. Sometimes the best instructional decision is not to use it at all.
Protect the Thinking That Matters
Productive struggle is still part of learning. Students need chances to wrestle with an idea, make an imperfect choice, realize that it does not quite work, and revise it. If AI enters too early or does too much, it can quietly become a substitute for that process.
ALIGN is built around a simple principle: AI can support the learning, but the student should remain responsible for the important decisions. The teacher should be able to identify the thinking that still belongs to the learner and explain why AI is helping rather than replacing it.
That principle also shapes how students use the tool. ALIGN includes a student routine called ROLE that gives the interaction more structure without turning AI into the author of the work. It asks students to clarify the task, define limits on AI’s role, and identify the kind of support they actually need.
The point is not to teach a clever prompt formula. Students still need to judge the response, decide the validity of advice, recognize when they are offloading too much, and decide when they do not need AI in the first place. That is a more useful definition of AI literacy than simply learning how to get better answers from a chatbot.
Teacher Judgment Needs the Same Protection
The same principle applies on the teacher side. AI can save educators time, but efficiency is not automatically the same thing as good teaching. If a feedback tool simply grades student work for us, we have moved the judgment away from the teacher instead of making the teacher more effective.
That is why one of the first ThinkTorch tools is the Teacher Feedback Assistant. The teacher supplies the assignment context and feedback criteria, and the system analyzes the student draft, identifies relevant evidence, and produces structured feedback and scoring recommendations for the teacher to review.
The distinction matters. The AI is not supposed to become the teacher of record. Its job is to reduce some of the mechanical workload so the teacher can spend more attention on what requires professional judgment: deciding what matters, whether the feedback is fair, and what the student most needs next.
For me, this is the same issue we face with student AI use. The question is not whether AI can complete part of the work. The question is whether giving that work to AI improves the human decision that follows.
What This Can Look Like in a Real Lesson
Imagine students are preparing an argument. They choose a position, develop a working claim, and select evidence. At that point, AI might take on a narrow role: act as a skeptical reader and raise questions the argument has not addressed.
The AI does not decide which criticism matters. The student does. A student can reject a weak suggestion, respond to a legitimate concern, revise the reasoning, and explain why the revision made the argument stronger. Later, the teacher can ask for an AI-free explanation, paragraph, conference, or other performance that makes the student’s independent understanding visible.
In that use, AI creates useful friction without becoming the author of the thinking. The student still has to evaluate, decide, revise, and explain.
Putting ALIGN Into Practice
ThinkTorch is where those ideas become usable. The platform is being built around tools that help teachers make the kinds of decisions ALIGN calls for, rather than simply using AI to generate more content faster.
The Teacher Feedback Assistant applies the same human-led philosophy to teacher feedback. “Should This Lesson Use AI?” helps educators decide whether AI belongs in a lesson at all. When it does, the ALIGN Lesson Designer helps teachers shape a purposeful role for it while keeping the important thinking with students.
The tools do different jobs, but the standard is the same: AI should strengthen the human thinking and judgment that follow.
We Do Not Have to Choose Between AI and Learning
Schools are often pushed toward two extremes: treat AI primarily as a threat and keep it out of the classroom, or treat it as inevitable progress and find a use for it everywhere.
Neither approach gives teachers enough credit. Educators already make decisions about when to scaffold, when to step in, when to step back, and when students need to struggle through something themselves. AI should be another instructional decision, not an exception to that judgment.
Students should leave school knowing how to use AI. They should also know how to work without it, question it, and recognize when using it would cost them the very practice they need.
That is the idea behind ThinkTorch: Think clearly. Learn deeply. Use AI wisely.
Interested in ThinkTorch AI?
ThinkTorch AI is preparing for early access. Educators, instructional leaders, and schools interested in trying the tools or following the development of ALIGN can request early access to ThinkTorch AI.
Request early access to ThinkTorch AI Here.
Stay in touch with NCCE by connecting with us! https://ncce.org/
About the Author
Christian Gross is a high school English teacher and department chair in Idaho, a former college composition instructor, and the founder of ThinkTorch AI. His work focuses on practical, human-led uses of artificial intelligence that support learning without replacing student thinking or teacher judgment.