Spark Session — Cognitive-Process Assessment in Practice

Hi, I'm Christie Shin. I teach UX/UI Design at FIT in New York. My challenge for this fellowship starts with a problem I see in my own classroom: once AI can produce a polished screen in seconds, the artifact stops proving who actually has the design expertise. So I'm building a way to grade the thinking behind the work, not just the work itself. I call it Cognitive-Process Assessment, CPA for short.

This isn't hypothetical. Before I built anything, I checked it against my own institution. FIT has no AI detection tools in place today. That conversation, with our Technical Manager for Faculty Research Space and Emerging Technology, told me grading the artifact was never going to work here. If the institution can't detect AI use, grading the process is the only path left. That's the premise CPA is built on.

Research starts human led: interviews, competitive analysis, a live site audit. Gemini Notebook synthesizes that into a brand profile and content brief. From there, students move into Gemini and Gemini Gems to write and direct UX copy, documenting their reasoning as they go. Google Stitch generates the first UI pass from an art direction the student already chose. Nano Banana fills in visual assets. Everything comes back into Figma, where the student applies real craft: typography, hierarchy, a working design system, catching whatever the AI got wrong along the way.

The part I grade is what happens at every one of those handoffs. Did the student catch an Attribution Error, a fact the AI just made up, or a Logical Error, a step that doesn't follow even from true premises? That language comes straight from Google's own REVEAL framework, which I connected to my rubric after Session 1. I track eight dimensions total, from prompt architecture to direct creative contribution, what the student actually sketched, wrote, or rewrote by hand. Each one has its own evidence trail: prompt logs, revision history, timestamped process notes.

The workflow already produces work that wins. CPA is the rubric that makes the thinking behind it legible and gradable.

Bunny Quest, by Kokona Akatsu and Haein Noh. They personified each of our CT&D minors as a bunny mascot, then extended that cast into posters, stickers, social personas, and an AI-powered quiz that matches students to a program. A range of AI tools produced the work. But the Gold Cube they won was in AI Process, not AI Output. The award was for how they directed it. That is exactly what CPA is built to grade.

CPA doesn't depend on a licensing tier or detection software, so it can work for any program facing this gap, not just the well resourced ones. If you're wrestling with the same question, how do you know a student actually did the thinking, I'd love to compare notes. I also share more case studies and insights from industry professionals in my book, Empowering Design Education with AI.

• Read the FIT quote in Slide 2 as a beat, not a rush. It's the strongest credibility line.
• Slide 3 is the densest. Let the flow diagram carry the tool names.
• Slide 5 numbers land as facts, not a sales pitch. Pause after "21 Young Ones awards."
• If running long, trim Slide 3 first, then the second half of Slide 5's stat list.