Academic AI Adoption & Assessment Impact Simulator

The academic assessment fixture assigns task weights forty five, twenty and thirty five, with risk scores eighty five, ninety two and twenty five. Weighted sum divided by one hundred equals sixty five point four. Yet the default computer science state overrides that arithmetic with a fixed baseline seventy eight point five whenever its first task weight is forty five. Bars span one hundred thirty point eight and one hundred fifty seven at two pixels per assigned score point. This is a calibration branch authored in the source, not a survey-derived vulnerability estimate or measured student behavior. Discipline adoption numbers and assessment risk labels are bundled assumptions with no empirical evidence established in this local run. The oral review policy multiplies baseline seventy eight point five by point two eight, then rounds to one decimal, yielding twenty two. Relative reduction is seventy two percent after rounding. The prohibition policy uses point eight eight, giving sixty nine point one and twelve percent relative reduction. Bars span sixty six and two hundred seven point three at three pixels per assigned remedied score. These are hypothetical multipliers, not a causal estimate that an actual policy will reduce misconduct by those percentages. The reliability caption is one hundred minus rounded remedied vulnerability plus ten point two, with a special cap branch. Recommendations can inform discussion, but these captions do not certify detector accuracy or institutional policy efficacy. Changing discipline from computer science to business removes the fixed default branch and combines point seven times weighted task risk with point three times discipline baseline seventy four. With the same task weights this gives sixty seven point nine eight, displayed sixty eight point zero. Bars span one hundred fifty seven and one hundred thirty six for default computer science and business captions at two pixels per displayed score. Grading reconstruction weights remain forty, thirty and thirty regardless of user edits. The renderer checks for an undefined D3 library, writes a Visualization Engine Loading placeholder and returns. Metrics, native controls and tables still update, and export listeners were installed beforehand. Actual JSON and brief downloads can therefore preserve the authored state despite a blank chart, without validating the empirical assumptions, exhaustive functions or public deployment.