Can local community apps/information systems be effectively infused with open AI to allow information to be personalised?
Client: Student project for a UX design short course at University of the Arts, London
Sector: Placemaking, open AI
Role: Research, conceptualisation, wireframe, design and prototype with Margarida Desport
Project time: 1 week
Stratford. was born out of primary and secondary research where some of the questions we were looking at were:
How can we use open AI to make smaller tasks easier for people commuting or discovering a place for the first time?
Can we present different depths of information to help someone discover activities and utilities in a new place?
Research methods
Direct storytelling
Direct storytelling is a narrative technique in which the storyteller communicates the story's events, characters, and emotions straightforwardly and clearly, without relying on complex structures, multiple layers of meaning, or heavy use of symbolism. We used this generate insights from the bottom up analysis (i.e. come up with insights based on participant data).
Our initial group concept was about simplifying complex information systems eg. local authority websites and how users interact with the open AI systems and creating an application based on principles rectifying possible errors in both these systems.
Our questions were: Which was the last online application you filled out, did you find it frustrating and why?
How often do you interact with ChatGPT? What kind of day to day activities do you use it for?
The interviews were each 20 minutes long.
Some of our findings from this research were:
Our participants used ChatGPT for admin tasks, and to find resources that might be reliable in an educational context but were hesitant to use it an ultimate source of truth. They also spoke about instances like this in which the information given by the open AI model was in fact factually incorrect.
Our participants pointed out that filling out complex information systems online, especially for an arduous filing processes like self assessment, it might be helpful to have language like ChatGPT which decodes the questions/requirements at each step of the process.
Contextual research (structured observation)
Contextual research is a method of studying and understanding people, behaviours, environments, or products by examining them within the context of their natural settings. It aims to gather insights that are deeply rooted in the real-world conditions and circumstances in which the subject exists. Since the campus was based in Stratford and our first branch of research was about easing complex information systems online we looked at doing this in Stratford station which was a rich intersection of the same but as a physical application. We looked at the A.E.I.O.U method i.e segregating our research by activities, environment, interactions, objects and users.*
Activity: Some commuters are confident and know where there are headed while others are look out for signs and symbols that will point them in the right direction.
Environment: Busy, not defined. The tapping of cards/tickets is a constant sound.
Interactions: Commuters were interacting with their phones/watches mostly and on certain cases with physical signs.
Objects: Clocks, signage, information booths, ticket machines.
Users: Two major categories: In transit/movement: passengers; stationary: TFL staff in vests
*There was a longer list which detailed all the observations made during this exercise.
Thematic coding
Based on the above, we created themes from the research which we would base the application on. Our two key themes were:
Establishing a reliable flow of information
Personalising information in a way that did not overstep privacy concerns
Creating wireframes based on Neilsen’s 10 usability heuristics
Our first wireframes were based on our key principles. The app would inform the user of the activities and options in Stratford, giving them different depths of information based on their requirements as well have an open API option to help create a more personalised flow of information like ‘How do I get to Stratford from my place?’
We looked at existing way findings apps like Citymapper and the TFL website as well as open AI interfaces like ChatGPT for analysis.
Based on the first wireframes and mocking them up on Figma we made participants test the app by asking them to complete a task on the app, in this case it was, ‘How do I get to Stratford?’ Based on these tests we made changes to the flow of how we wanted to direct the user, the ask me anything app had different levels of information shown like “look up something” and “talk to us”. We tried to ascertain that these two would give them the most reliable information as they would be seen by the user first.
Apart from way finding, the app also looked at events and activities happening around Stratford, for example offers in restaurants and shops at Westfield mall as well as sporting events at the stadium. The app would help bring together the community aspect of Stratford while relying on the key principles of reliability and personalisation. This could also be scaled to use in different local authority websites and applications.
Some key features of the app were:
We presented our research and wireframes at the end of the week for feedback.
