Prism
With emerging AI technologies, giving and managing informed consent online is a formidable task.
Prism aims to make it a little easier.
My role
This was a solo project and I worked on all aspects.
Figma, Excel, Survey writing, Data Analysis, Thematic Analysis, Literature Review, Report Writing, Research Design, Research Presentation, Premiere Pro
Results
In 12 weeks, I designed and conducted a research project into data practices in Recommender Systems.
This culminated in Prism, a vertical slice through a digital product which responded to the research.
Prism is designed specifically for people with low to middling digital literacy and provides an easy-to-use VPN, Adblocker, recommender system (RS) and education hub.
By designing Prism, I aimed to create a ‘front page of the internet’ where everyone, regardless of their opinions, could recieve an enhanced service.
User Insights
Using value sensitive design, I created a simple prototype in Figma. I focussed on the Recommendation Flow, from discovery to customisation.
Users can ‘tweak the algorithm’ to control their personalised recommendations. Content is collated on the dashboard which is not only customisable, but fully controlled by the user.
Simple browsing controls allow users who don’t care about privacy to still receive an enhanced digital experience.
Varied levels of control and privacy respond specifically to user feedback. Some participants were very worried about data collection, while others just wanted more relevant content.
Further development of the overall accessibility is an important next step, especially regarding the dashboard.
Reflection
Creating Prism made me rethink what user-centric design means.
User responses to my survey were similar but varied in many subtle but important ways. People were angry, scared, happy, disinterested. By keeping that in mind, I created a product that they can all benefit from.
Reflecting on Prism, I think it successfully gives the user behind-the-scenes information on how they recieve and control their recommendations. This allows them to provide continuous, enthusiastic and evolving consent as standard.
I would love to develop the algorithm controls further to prevent self-induced echo chamers while still retaining the balance of giving the user full autonomy.