An AI Suggestor to Help Instructor in University of Tasmania Create Innovative Assessment Strategies
ROLE
DESIGN LEAD FOR CAPSTONE PROJECT
MISSION
PROVIDE ASSESSMENT SUGGESTIONS LEVERAGING AI
TIMELINE
JAN 2024 - AUG 2024
BACKGROUND
As generative AI become more prevalent, instructors in University of Tasmania face challenges in designing assessments that evaluate authentic learning. If traditional forms of assessments like essays and exams can no longer assess learning, then -
How should we facilitate instructors in University of Tasmania in creating innovative types of assessments that accommodate the advent of AI?
PRODUCT PREVIEW
RESEARCH
USER INTERVIEW
To dig into the problem, we started with an user interview with multiple professors from two colleges (College of Arts, Law, and Education and College of Health and Medicine) in UTas to learn about their needs and goals.
We found that our users value assessing higher-order thinking and are excited about the possibility of AI, but caustiousness around the integrity of AI-driven assessment still exist.
.png)
COMPETITOR ANALYSIS
Existing tools help create assessments by generating questions like multiple choice or fill-in-the-blanks, but they mainly focus on assessing memorization.
None target higher-level skillset such as applying, analyzing, or creating, or consider the role of AI in the assessment design. Our product aims to give the instructor both guidance and freedom in designing both AI-assisted or AI proof assessments.
.png)
USER PERSONAS
According to the user interview, we identified common pain points and needs based on the different disciplines those professors are teaching.
JOURNEY MAP
RESEARCH TAKEAWAYS
DESIGN
MVP 1
From our ideation process, we decided to move forward with the concept of an AI assessment type generator. We drafted out MVP1, which is a simple Web App chatbot connected to a customized GPT Assistant API.
CONCEPT VALIDATION
We performed think-aloud usability testing with 6 professors on our MVP1 and got some great feedback related to the experience of interacting with the chatbot.
Many users do not know how to prompt the AI to get their desired outcome. They hope to have a scaffolded, guided approach of prompting, with AssessMate suggesting what kind of information they should input. Also, they hope to have a feature that can customize the degree of AI usage in the assessment.
FINAL DESIGN
Based on the feedback we received from MVP1, our final design visualizes and optimizes the user experience, offering more guidance and scaffold for the user to reach the desired assessment outcomes.
UI DESIGN
COLORS
TEXT COLORS
LOGO & BUTTON
FONTS
ONBOARDING
To start with, the user selects their goals of using AssessMate. The design decision is that user choose between a “step-by-step” guiding interaction and a free chatting interaction. This gives users the autonomy to select whether they want a guided approach or free prompting.
01
communication
There is nothing more important than communication for any project. Communicate with teammates, communicate with stakeholders/clients, communicate with developers - constant and smooth communication is the key to a great end-product.
02
shitty first draft
Don't be afraid of shitty first draft. The minimal viable product or our first baby might not be perfect in features, but we can keep iterating and plan future steps.
03
small detail matter
I realized how minor change in design decision can lead to significant shift in user experience and satisfaction.
OTHER PROJECTS
