A HARVEST KERNEL WORKING SESSION
The Evidence Map
Bring one syllabus. Take the three course outcomes that matter most, list the graded work that counts as evidence for each, and mark which of that work you actually watched being done. Then flag the outcomes that rest only on work nobody saw, stress test one assignment yourself, and choose one change for next term. You leave holding a one page evidence map you can show a chair, a colleague or a review committee. You do the judging. Nothing here decides it for you, and nothing you type leaves this page.
Why this exists. On November 11, 2026, England’s higher education regulator, the Office for Students, published a report with Advance HE on AI in teaching and assessment. The views of 244 staff and 91 students and student representatives from at least 127 institutions informed it. Its rule for assessment is short: “If it becomes possible to complete an assessment using AI without the knowledge and skills being tested, the assessment should be replaced.” It adds that assessing knowledge, rather than skills, “will usually require assessments to be secure.” The participants were self selecting, and this session is a judgment call built from one report, not a tested method.
SEED
Bring the syllabus
Open the syllabus you will teach next term. Write the course name, and how many different courses you teach in a year. Then pick the three learning outcomes that carry the most weight and copy each one word for word. Not a paraphrase. The exact promise your syllabus makes.
SPROUT
Attach the evidence, and mark what you watched
For each outcome, list every graded piece of work that counts toward it. Then, in the second box, list only the pieces you or a colleague saw happen: an in class write, a presentation, a conversation about the work, a lab, a performance. Anything done anywhere, with any tool, and turned in as a file does not go in the second box. If nothing was watched, leave it empty. Last, mark the outcome as knowledge, meaning knowing the concepts of the field, or skill, meaning doing something with them.
GROW
Flag the gaps, then stress test one
Press the button. Any outcome with nothing in its watched box gets a flag, and a flagged knowledge outcome goes to the top of the list. Then take the assignment behind your top flag, paste your own prompt into any chatbot, and read what comes back as if a student had turned it in. The report describes an institution that stress tested its assessments exactly this way, by having staff try to complete them with AI tools.
Your flags will appear here.
HARVEST
Choose one change for next term
The report groups the redesigns it found into three families. A secure piece: a short conversation about the submitted work, an in class write, a presentation. A process piece: a checkpoint graded before the final draft. Or an AI piece: students critique AI output, and you grade the critique. Pick one family for your top flag, press the button for a starter sentence, and rewrite it so it sounds like you.
