Why This Exists

You cannot teach AI literacy without having AI in the room. This tool is built on a single research-backed principle: friction by design. AI is the most efficient discomfort-eliminator ever built — which is precisely why we have to engineer the struggle back in. Research consistently shows that retrieval, productive struggle, and the discomfort of not-yet-knowing are where understanding takes hold. AI undermines learning only when used as a substitute for thinking on problems within a student’s reach. The reflection checkpoints, the visible choices, the report-out — these are not administrative steps. They are the friction. Studies of human-AI collaboration consistently find that a small minority of students who use AI as a sparring partner — demanding evidence, asking for counterarguments, refusing to settle — outperform those who simply accept its output. This tool is built to produce more of those students.

The gap between receiving a question and responding to it is where thinking actually happens. AI closes that gap by design — it produces output the moment input arrives. Instructional design that preserves thinking has to build the gap back in deliberately: students produce something first, on their own, before the AI enters. What they wrote before the tool arrived is the evidence that thinking happened. What they changed afterward — and why — is the learning.

The BCcampus GenAI in Teaching and Learning Toolkit (Nguyen, 2024) offers a five-level AI Assessment Scale — from No AI through Full AI as co-pilot — that maps directly to how this tool structures student choice. The Classroom Session Builder is designed to operate at levels 2 and 3: AI-assisted thinking, with student judgment governing what gets accepted, challenged, or discarded. No AI content enters the final output without passing through a human decision. The process is the assessment. The product is evidence of it.

Academic integrity is not a technical problem. Detection tools fail on hybrid work — which is almost all real student work — and research shows they flag non-native English speakers at rates approaching 61–98%. The answer is not better detection. It is assessment designed so that the thinking cannot be outsourced: process documentation, visible choices, oral report-outs, reflection checkpoints. This tool is built on that premise.