Two scores, not one
The evidence file separates the product-quality score from the understanding score, shows the gap between them, and suggests questions for a short conversation before approval.
Edython 2026 · Smart Learning and Education Quality track
From digital product to proof: assessment that reveals real understanding.
Burhan generates questions from the student's own product to separate product quality from understanding.
From a fictional student's evidence file in the prototype
A Team 3 project at the Edython forum for innovative educational solutions, College of Education, King Abdulaziz University
The Edython challenge
Today a student can produce a polished video or presentation in minutes with AI tools. The track's challenge asks: how do we make sure the student actually learned, and that the product reflects their own understanding and skills, not just the AI tool's abilities?
Burhan does not ban the tools or rely on detecting generated text. It allows the AI tool, asks the student to disclose its role, then gathers evidence of the student's own understanding.
How it works
Three steps: Burhan reads the lesson and the product, and the teacher makes the call.
Sets the goal; Burhan suggests.
Makes and discloses; Burhan reads.
Answers and fixes; the teacher decides.
The teacher decides
The evidence file separates the product-quality score from the understanding score, shows the gap between them, and suggests questions for a short conversation before approval.
The student states who made each part: themself, the AI tool, or both, and whether they checked it. The card is not graded.
They quote the student's own product, and each card is tied to a success criterion. They never reveal the error; they ask the student to justify, transfer, and verify.
Submissions sorted by gap. The teacher talks with wide-gap students and a random sample, and approves the rest in one go.
On the student's own device, not school machines. The teacher sees the page-leave log when deciding.
Burhan's journey
Three parties in one journey: the teacher sets the goal and decides, the student makes and proves understanding, and Burhan reads and suggests.
Burhan reads the textbook pages and extracts the objectives and success criteria; the teacher edits and approves them.
It suggests a digital product that fits the lesson activity and writes its specs; the teacher approves the assignment.
They make the product with any AI tool they choose, and the disclosure card shows the AI tool's role in each part. Disclosure is not graded.
It finds gaps between the product and the goal, writes three cards from the product itself, and puts first what was credited to the AI tool without checking.
“Why?” justifies a choice, “What if?” transfers understanding to a new situation, and “Verify” checks what the AI tool wrote. On their own device, with page exits logged.
A product score, an understanding score and the gap between them, a line on their thinking for each card, and a list of gaps.
They revise the product themselves and tick each gap they fixed; Burhan names the gaps but never designs for them, then checks the fixes.
The class board is sorted by gap. The teacher talks with those who have a gap and a random sample, keeps, raises or lowers the score, and approves the rest at once.
What is it built on?
Builds assessment on three linked models:
Mislevy, R. J., Steinberg, L. S., & Almond, R. G. (2003). On the structure of educational assessments. Measurement: Interdisciplinary Research and Perspectives, 1(1), 3–62. https://doi.org/10.1207/S15366359MEA0101_02
Effective feedback answers three questions:
Hattie, J., & Timperley, H. (2007). The power of feedback. Review of Educational Research, 77(1), 81–112. https://doi.org/10.3102/003465430298487
Team
Smart Learning and Education Quality track, Edython 2026.
Pick a lesson as a teacher, then hand in its product as a student, and try the cards in locked mode.