GPT for Communication: How can GPT Bridge the Communication Gap between Students and Universities?
ROLE: SOLE USER RESEARCHER AND UX DESIGNER
JUNE 2023 - AUG 2023
SKILLS: USER RESEARCH, USER-CENTRED DESIGN, QUALITATIVE RESEARCH AND ANALYSIS, CO-DESIGN WORKSHOP, EXPERT (HEURISTICS) EVALUATION, FIGMA, COMMUNICATION
As part of my Master's degree in Human-Computer Interaction (HCI), I took on this project for my dissertation. This topic resonated with me after working as a Research Assistant at the university, which inspired me to delve deeper into the challenges students encounter in their communication and interaction with the university.
Over the course of this three-month research venture, I had the opportunity to refine my skills in areas such as qualitative research, user research, user experience design, data collection and analysis, utilizing Figma, conducting expert evaluation testing, and, of course, effective communication.
This project aims to explore how Generative Pre-Trained Transformers (GPT) can be utilised as an intelligent communication tool to facilitate effective interaction between students and universities.
Background & Motivation
University student support services play a pivotal role in students' performance and success, and having access to timely, accurate information and personalized assistance stands out as one of the most vital.
Communication gaps and lack of student support pose issues such as:
1. Student dissatisfaction and frustration (Santoso, 2018, Waddington, 2010, Ranoliya, 2017)
2. Disconnection and negative perception of university (Gray, 2021)
3. Fragmented information dispersal and critical information gaps (Waddington, 2010)
"Communication" in this context:
Communication here refers to both academic and non-academic matters, including but not limited to:
-
Access to course and student support service information,
-
Well-being support,
-
Financial assistance,
-
Application guidance,
-
Educational support,
and any other student inquiries.
Why GPT?
Recent studies have explored AI integration into higher education institutions and found that it:
-
Enhances efficiency and efficacy (Rasul 2023)
-
Streamlines tasks, saves time and cost (Rasul 2023, Merelo 2022)
-
Reduces staff workload, enhance student engagement and satisfaction (Santoso 2018)
Research Questions
What are the specific communication challenges faced by staff and students in universities?
How can GPT-based systems be designed to address those challenges effectively and what are some of the benefits and drawbacks of these systems?
What are the key factors and features of the chatbot system that contribute to the successful adoption and implementation of GPT-based communication systems in universities?
Research Method
Research Phases
The research was divided into four distinct phases. Despite the demanding timeline of less than three months, I eagerly embraced the opportunity to explore the various HCI methodologies that were relevant to my research.
Participants Recruitment
Interviews: Purposive Sampling was employed to recruit 2 students and 1 staff member for the interviews. This approach ensured a comprehensive understanding of the challenges faced, gathering insights from both student and staff perspectives, key stakeholder groups within the university.
Co-design workshop: 6 university students were recruited through Snowball Sampling, a method ideal for projects with limited time, budget and those that involve discussing potentially sensitive topics.
Expert evaluation: The Expert Evaluation was conducted with a software enginner, recruited through Purposive Sampling.
Prior to the research, written and verbal consent was obtained from all participants. Each participant received an Information Sheet and had ample opportunity to ask questions for clarification.
Evaluation & Findings
Summary of Research Process
Phase 1: Semi-structured interviews
First, qualitative interviews were conducted with students and staff, representing key university stakeholders. Each interview, lasting 45 to 60 minutes, provided valuable perspectives on information challenges and AI tools as a potential solution.
The data was analysed using Braun and Clarke's thematic analysis method, resulting in five central themes.
Personas
Using the interview data, two semi-fictional personas were crafted: Alex and Sarah. These personas embody potential users who could benefit from an AI tool.
In the co-design workshop, they serve as discussion prompts and a foundation for understanding user needs prior to ideating the chatbot prototype.
USER STORY
As an undergraduate student, Alex wants to receive timely information and personalised recommendations in order to stay informed and make the most out of his university experience.
USER STORY
As an administrative staff member in Student Services, Sarah wants to employ an AI chatbot to efficiently handle common student questions while striking a balance between human intervention and AI and ultimately aiming to streamline her workload and enhance the quality of student interactions.
Phase 2: Co-Design Workshop (Ideation)
The 90-minute in-person co-design workshop involved six student participants and comprised two distinct activities.
Activity 1: Empathizing with Personas
Participants engaged with the personas and addressed their pain points, opportunities, and suggested features for a potential AI tool tailored to their needs.
RESULTS
Activity 1 in the co-design workshop validated Phase 1 findings from the interviews, reinforcing the 5 central themes. These insights were synthesized into a chart, providing empirical data for the research.
User Journey Map
Due to time constraints, the following segments (User journey map, user flow, and storyboard) were not included in the final report. However, I decided to develop them on my own time because they provide essential visual aids that enhance the comprehension of the user's interaction with the AI chatbot. The user journey map offers a holistic view of Alex's experience, showcasing key touchpoints and potential pain points. The user flow outlines the step-by-step progression of interactions, facilitating a clear understanding of the process while the storyboard visually narrates specific interactions.
Persona: Alex
Goal: To find out how to register for the Language Resource Centre and where to find part-time work opportunities.
User Flow
Storyboard
Activity 2: Co-Designing AI Chatbot Prototype
Afterward, participants received a Design Brief and were guided by the researcher to create a preliminary sketch of an conceptual AI chatbot. Examples were provided for reference and inspiration.
RESULTS
Following this, we created a set of AI chatbot prototype sketches. Simultaneously, we discussed and developed a list of crucial features, such as accessibility options, language translation tools, and the ability to save and email conversations. This was compiled into a comprehensive Design Guideline document.
Phase 3: Prototype Design
Using the sketches and the list of features derived from the co-design workshop, I constructed a mockup of the envisioned AI chatbot tool, using the Figma software.
A screenshot of Newcastle University's landing page was used in the mockup.
I also designed an expanded version of the AI chatbot tool that would take up the full screen, tailored for current users within the university system, such as existing students and staff members.
Lo-fi wireframe:
Hi-fi wireframe:
Phase 4: Expert Evaluation (Nielson's 10 Heuristics Guidelines)
To validate the design, I applied Jakob Nielsen's 10 Heuristics Guidelines and conducted the evaluation with an industry expert well-versed in both the heuristics and UX design. This expert is a seasoned software engineer based in South East Asia.
The assessment was conducted through a Zoom call. I facilitated access for the expert to review my prototype on Figma, and also furnished them with a user-friendly, fillable form based on the Heuristics Guidelines template.
RESULTS
The Heuristics Evaluation yielded insightful observations aligned with recognized usability heuristics. These findings reveal key areas demanding attention to enhance the system's usability and user experience.
%20(1)_.jpg)
LIMITATIONS
Given the project's time constraints and resource limitations, a fully functional chatbot prototype wasn't feasible. As a result, some usability features couldn't be demonstrated adequately. Nonetheless, I opted for an Expert Evaluation to gather valuable insights for future design iterations and to gain hands-on experience applying Nielsen's 10 Heuristics guidelines. Previously, my evaluations had centred on Cooperative Evaluation reports using the think-aloud protocol. I welcomed the opportunity to challenge myself with a new approach.
Conclusion
Research Questions: Answered ✅
Drawing on the comprehensive data collected across four distinct research phases, this research has illuminated several key insights which answer the research questions and delivered an artefact in the form of a prototype of the conceptual AI chatbot tool in this context.
Furthermore, wider implications have been identified and listed below.
For future work, it is hoped that this continued exploration into the boundless potential of AI tools such as GPT will persist. This research project, at the very least, seeks to kindle and propel future initiatives within this domain.