Assignments that are resilient to artificial intelligence do not require advanced technical expertise, they require intentional, active design. This is where the ‘RAD’ framework comes into play. RAD stands for Revise, Analyze/Apply, and Defend. By integrating these three core elements into existing assignments, teachers can build cognitive ‘speed bumps’ that prevent passive AI completion. The goal is to make student thinking the central focus of the assignment output, ensuring that even if a student uses AI as a supportive starting partner, they must still uniquely own and demonstrate their learning to earn academic credit.
The first element, Revise, requires students to show how their thinking changed over time. Instead of merely submitting a final, polished draft, students must reveal the progression of their ideas. For example, a teacher can ask students to highlight a specific sentence in their work where their understanding shifted, and write a brief note explaining what caused that change. This could include comparing an initial rough draft with a final version, or detailing how feedback from a peer or an AI tool helped them refine their argument. By highlighting these revision choices, students actively practice metacognition, and teachers gain a clear, documented window into the student’s learning journey and growth.
The second element, Analyze and Apply, pushes students to take abstract concepts and transfer them into unique, personalized contexts. AI can easily generate generic, textbook-style definitions, but it cannot replicate personal narratives, localized details, or lived human experiences. If a history assignment requires students to apply a constitutional amendment to a recent event in their own neighborhood, or connect a literary theme to a personal family story, the student is forced to actively synthesize the information. This application layer not only makes AI copy-pasting impossible but also makes the learning deeply personal and relevant to the student’s real life, promoting long-term retention of core standards.
The final element, Defend, challenges students to justify their academic choices and evidence. If a student turns in a polished product, they must be prepared to verbally or textually defend their work. This can be accomplished through simple written prompts, such as: ‘Why did you select this specific quote over others?’ or ‘Why is this source trustworthy to you?’ In class, this can be reinforced through quick ‘turn-and-talk’ sessions, peer reviews, or Socratic seminars. Getting RAD with assignment design does not make the workload bigger for teachers or students; rather, it makes the expectations clear, ensuring that human thinking remains the non-negotiable prerequisite for earning a grade.
By implementing the RAD framework across different subject areas, schools can create a cohesive instructional language. A science teacher can ask students to defend their lab results, while a math teacher can require them to explain where they corrected a calculation error. This framework turns AI from a threat into an asset, allowing students to use technology to expand their thinking, while requiring them to use their own beautiful minds to demonstrate their actual understanding, ensuring they are actively engaged in self-monitoring and cognitive ownership from start to finish.
NEXT STEPS AND RESOURCES
Watch the Session 2 video presentation, ‘What Learning Evidence Should Still Show,’ on the Illuminate XR YouTube channel and get in touch with the IXR team at illuminatexr.com/handbook to see how we can help your staff design effective assignment expectations. Contact us to collaborate on updating your school’s grading and expectation guidelines.
Assignments that are resilient to artificial intelligence do not require advanced technical expertise, they require intentional, active design. This is where the ‘RAD’ framework comes into play. RAD stands for Revise, Analyze/Apply, and Defend. By integrating these three core elements into existing assignments, teachers can build cognitive ‘speed bumps’ that prevent passive AI completion. The goal is to make student thinking the central focus of the assignment output, ensuring that even if a student uses AI as a supportive starting partner, they must still uniquely own and demonstrate their learning to earn academic credit.
The first element, Revise, requires students to show how their thinking changed over time. Instead of merely submitting a final, polished draft, students must reveal the progression of their ideas. For example, a teacher can ask students to highlight a specific sentence in their work where their understanding shifted, and write a brief note explaining what caused that change. This could include comparing an initial rough draft with a final version, or detailing how feedback from a peer or an AI tool helped them refine their argument. By highlighting these revision choices, students actively practice metacognition, and teachers gain a clear, documented window into the student’s learning journey and growth.
The second element, Analyze and Apply, pushes students to take abstract concepts and transfer them into unique, personalized contexts. AI can easily generate generic, textbook-style definitions, but it cannot replicate personal narratives, localized details, or lived human experiences. If a history assignment requires students to apply a constitutional amendment to a recent event in their own neighborhood, or connect a literary theme to a personal family story, the student is forced to actively synthesize the information. This application layer not only makes AI copy-pasting impossible but also makes the learning deeply personal and relevant to the student’s real life, promoting long-term retention of core standards.
The final element, Defend, challenges students to justify their academic choices and evidence. If a student turns in a polished product, they must be prepared to verbally or textually defend their work. This can be accomplished through simple written prompts, such as: ‘Why did you select this specific quote over others?’ or ‘Why is this source trustworthy to you?’ In class, this can be reinforced through quick ‘turn-and-talk’ sessions, peer reviews, or Socratic seminars. Getting RAD with assignment design does not make the workload bigger for teachers or students; rather, it makes the expectations clear, ensuring that human thinking remains the non-negotiable prerequisite for earning a grade.
By implementing the RAD framework across different subject areas, schools can create a cohesive instructional language. A science teacher can ask students to defend their lab results, while a math teacher can require them to explain where they corrected a calculation error. This framework turns AI from a threat into an asset, allowing students to use technology to expand their thinking, while requiring them to use their own beautiful minds to demonstrate their actual understanding, ensuring they are actively engaged in self-monitoring and cognitive ownership from start to finish.
NEXT STEPS AND RESOURCES
Watch the Session 2 video presentation, ‘What Learning Evidence Should Still Show,’ on the Illuminate XR YouTube channel and get in touch with the IXR team at illuminatexr.com/handbook to see how we can help your staff design effective assignment expectations. Contact us to collaborate on updating your school’s grading and expectation guidelines.
Accusing a student of academic dishonesty based on unstable software detectors is a rapid path to destroying classroom trust. When a teacher receives a paper that seems too polished, too…
The straight-A graduate who cannot navigate an unstructured workspace is one of modern education’s most silent crises. This represents a systemic, growing chasm between the design of public education and…
The fear that new technology will erode student intelligence is a historical cycle, not a modern phenomenon. They argued that if a machine could write an essay, solve a math…
Vague guidelines and inconsistent classroom rules leave students confused and anxious about when and how they can ethically use artificial intelligence. When policies are vague, overly punitive, or vary wildly…
