How to Grade Student Work When AI Can Write It

Grading student work got harder the moment AI could write it. Mike Jones has a fix, and it starts with grading something other than the finished artifact.Most courses still assess a thing: the essay, the spreadsheet, the safety manual. AI produces things on demand, so faculty end up policing with detection software Mike says the data shows is unreliable.WHAT YOU'LL TAKE AWAYWhy the artifact stopped being evidence of learningThe five AI modes: hone, use, mediate, align, navigateThe second rubric, grading how students used the toolBoth equity gaps: banning AI, and saying "use whatever you want"Why a $20 subscription is not FERPA-safeTHE ASSESSMENT FIXMike's HUMAN framework gives five operational modes for AI in coursework: hone, use, mediate, align, and navigate. The first is hone, meaning no AI, and it exists because students have to build a foundation before they can audit what a machine hands them.His example is a workplace safety manual. AI will write you a good first draft. If you cannot explain how a forklift operates, you cannot check that draft, and you are still the human legally holding the outcome when someone gets hurt.HEART is the backward design side, producing two rubrics on the same assignment: one grading the demonstration of learning, one grading intentional use of the tool. The assignment often does not change. What changes is what you look at when you grade it.BOTH EQUITY GAPSBlock AI entirely and your students graduate into a workforce fluent in something they never touched. Say use whatever AI you want and one student brings a $100 ChatGPT account while another uses the free tier, so you built a gap trying to be generous.Mike also points to peer-reviewed research showing generative AI helping students with ADHD and dyslexia. His concern is what a blanket ban does next: it forces students to disclose a disability to earn an exception, then invites faculty to litigate whether the diagnosis is real.THE TAKEAWAYIf AI can produce the artifact, the artifact is no longer the evidence, so assessment has to move to the process a student can show and defend.YOUR CHALLENGETake one assignment you already teach. Label its AI level in the instructions, in plain language. Then write the second rubric that grades how the student used the tool.ABOUT MIKE JONESAI literacy and onboarding specialist, co-founder of AI Engines in Marion, Indiana, and creator of the HUMAN/HEART AI-Resilience Framework. 15 years in higher ed teaching and learning technology.LinkedIn: https://www.linkedin.com/in/mediamission/AI Engines: https://aiengines.work/RESOURCES MENTIONEDAI Engines services and HUMAN/HEART → https://aiengines.work/servicesHECC Conference, Indianapolis, November 2026 → https://hecc.k12.in.us/conference-info.htmlCONNECT WITH JASON PADGETTJason Padgett | AI Solutions Architect | Co-Founder, Wooden Boy Industries | Host, AGI Podcast | AI Adoption, LLM Systems & Human-AI WorkflowsAGI - Advance Grow and Innovate with AI Podcast Network 🌐 https://schoolsout.agipodcast.ai/CoFounder, Wooden Boy Industries 🌐 https://woodenboyindustriesai.com/Chief Solutions Architect, Chakra AI Solutions 🌐 https://chakrasolutions.ai/📧 [email protected] FOR INTERVIEWS AND KEYNOTES ON AI IN [email protected] PODCAST SHOWSSchool's Out Saturdays: AI in education, K-12 through universityWeekly Blitz: AI news every MondayThe Vibe: Deep conversations, no clockAll platforms: https://linktr.ee/AGIPodcastFollow: Facebook + LinkedIn @AGIPodcast | Instagram @AGISchoolsout | X @NoCodeCAIOAI in education events around Indiana are listed at https://schoolsout.agipodcast.ai/TOOLS JASON RECOMMENDSRiverside.fm → https://riverside.sjv.io/e1xyBrWispr Flow → https://ref.wisprflow.ai/jason-padgett