By Dr. Janette Camacho | July 22, 2024
By the time you read this, your students have already used AI today. They used it before first period, probably before breakfast. The question facing educators in the summer of 2024 is not whether AI belongs in schools - it is there - but how we govern its use in ways that protect students, preserve academic integrity, and prepare young people for an AI-saturated world.
The problem with most AI use frameworks I have encountered over the past year is that they were written by committees optimizing for legal defensibility rather than classroom usability. They tend to be long, abstract, and focused on prohibition. A policy that says "AI tools may be used only with prior teacher approval for specific assignments as documented in the lesson plan" is technically precise and practically useless for a teacher managing 150 students across five periods.
After working with educators in over 15 schools during the 2023-2024 academic year - through my role as a Google for Education Certified Trainer and through direct consulting with district leadership teams - I have developed a framework I call CLEAR: Cite, Limit, Evaluate, Attribute, Reflect. It is not perfect, but it has the virtue of being memorable, teachable, and adaptable across grade levels and content areas.
The CLEAR Framework
C - Cite Your AI Use
The simplest and most powerful norm I have introduced is requiring students to cite their AI interactions just as they would cite any other source. This does not mean writing a formal MLA citation for every ChatGPT prompt (though at the high school level, I do require an appendix of significant prompts). At the elementary level, it can be as simple as a checkbox on the assignment: "I used an AI tool to help with this work: Yes / No. If yes, I used it for: [brainstorming / editing / research / other]."
The pedagogical principle here is transparency, not punishment. When students learn to disclose AI use as a routine practice, it normalizes the tool while making the teacher-student conversation about learning processes possible. I have found that citation requirements actually increase responsible use - students think more carefully about when and how they engage AI when they know they will need to articulate that engagement.
L - Limit the Scope
Every AI-integrated assignment should specify the boundaries of acceptable AI use before students begin. I provide what I call a "green light / yellow light / red light" guide for each major assignment.
Green light activities are those where AI use is encouraged: brainstorming topics, generating outlines, finding counterarguments to stress-test a thesis, translating concepts into simpler language for comprehension checks.
Yellow light activities require teacher approval: using AI to draft portions of text that will be revised, employing AI for data analysis that the student must then interpret and explain, generating code that the student must debug and annotate.
Red light activities are those where AI use undermines the learning objective: taking a math assessment where the goal is demonstrating procedural fluency, writing a personal narrative where voice and lived experience are the point, completing a lab observation that requires direct sensory engagement with materials.
The key insight is that limits should be assignment-specific, not blanket policies. An AI tool that is a red light for a timed writing assessment might be a green light for the brainstorming session that precedes it.
E - Evaluate the Output
This is the critical thinking component, and I consider it the most educationally valuable element of the framework. Students must evaluate AI-generated content before using it. I teach a structured evaluation protocol with four questions:
- Is it accurate? Verify factual claims against authoritative sources. AI confidently produces fabricated citations, incorrect dates, and plausible-sounding misinformation.
- Is it complete? Does the AI output address the full scope of the question, or has it oversimplified or omitted important nuances?
- Is it biased? What perspectives are centered? What voices are absent? Whose experience is treated as default?
- Is it mine? After evaluation and revision, does the final product reflect my understanding, my argument, my voice?
Teaching students to interrogate AI output is arguably more valuable than any content standard I cover. It develops exactly the kind of epistemic vigilance that citizens need in an information environment increasingly shaped by generative AI.
A - Attribute the Human Work
This principle reverses the typical framing. Instead of asking "how much did AI do?", I ask students to explicitly identify and take ownership of their human contributions. What decisions did you make? What did you change from the AI suggestion, and why? What knowledge or experience did you bring that the AI could not?
In practice, I implement this through reflection prompts embedded in the assignment rubric. A student who uses AI to generate a first draft of a persuasive essay must submit annotated revisions showing where they departed from the AI output and explaining their reasoning. The annotation is often where the deepest learning becomes visible.
R - Reflect on the Process
The final component is metacognitive. After completing an AI-assisted assignment, students write a brief reflection addressing: What did AI do well? Where did it fall short? How did using AI change your thinking process? Would you use it the same way next time?
These reflections have been revelatory. Students frequently report that AI was most useful for overcoming "blank page paralysis" and least useful when they needed to express personal conviction or creative originality. Many develop increasingly sophisticated prompt engineering skills over the course of a semester - not because I teach prompting as a discrete skill, but because the reflection process naturally drives them to iterate on their approach.
Implementation Lessons
After piloting CLEAR in 15 schools across three districts during spring 2024, several patterns emerged.
Start with teacher buy-in, not student compliance. Teachers who experienced the framework as learners first - using AI for their own planning and applying CLEAR to their own process - implemented it with students far more effectively than those who received it as a policy directive.
Elementary adaptation is possible and necessary. Third graders can learn to cite AI use and evaluate output at a developmentally appropriate level. I observed a third-grade teacher who had students use AI to generate three possible story endings, then vote on which was "most like a real kid wrote it" and discuss why. That is evaluation in action.
The framework reduces academic integrity violations. Two of the three pilot districts reported a decrease in AI-related academic integrity referrals after implementing CLEAR. When you give students a legitimate, structured pathway to use AI, most of them will take it rather than using it covertly.
It evolves with the technology. When Google released Gemini Advanced and OpenAI launched GPT-4o during the pilot period, the framework did not break. CLEAR is tool-agnostic by design - it governs the relationship between human and AI output regardless of which specific platform generates that output.
The Bigger Picture
I developed this framework because I believe that responsible AI use is a teachable competency, not an innate character trait. We do not expect students to intuitively know how to evaluate a website's credibility - we teach them CRAAP tests and lateral reading strategies. AI literacy deserves the same instructional intentionality.
The districts that are getting this right in 2024 are the ones that treat AI policy as a pedagogical opportunity rather than a compliance burden. They are asking not "how do we prevent misuse?" but "how do we develop the judgment that makes misuse less likely?"
CLEAR is one answer. It is not the only answer, and I expect to revise it as the technology and our collective understanding evolve. But it is a starting point that works in actual classrooms with actual students - and in the current landscape, that matters more than theoretical elegance.
Dr. Janette Camacho is a Google for Education Certified Trainer & Coach, Google Certified Educator Level 1 & 2, Adobe Creative Educator, Apple Teacher, and FETC 2024 Featured Presenter with 28+ years of K-12 classroom experience. She is the founder of iTeachAI.