︎




Facebook: The Hidden Bias in Newsfeed Algorithms


The Center for Applied AI (CAAI) discovered that the facebook newsfeed algorithm penalizes out-group content more than it should due to rushed user behavior. The purpose of this project was to clearly present these findings to a digital audience and teach them to be mindful when scrolling their newsfeeds.  


Overview

My Role: Visual Designer, UX Designer, UX Researcher, UI Designer
Team: Florence Ukeni (writer), Emily Bembeneck (director) 
Timeline: 6 months 
Tools: Figma, Adobe Illustrator 
Deliverables: User research, User Flows and Information Architecture, Wireframes, Illustrations and Data Visualizations

Design process






User Research 

The goals of our user research were to characterize our audience, determine how to best address them, and strategize which content should be emphasized in the user flow. We worked to answer the following questions: 
  • What does our primary audience look like?
  • What does our audience know about AI and the facebook algorithm? How do they feel about how their social media feed prioritizes content? 
  • What does existing research say about user reactions to social media algorithms? 
  • Which content from the research paper best supports our goals?

Methodology

  1. Conduct Primary Research: polling and peer interviews
  2. Conduct Secondary research: online research and competitor analysis
  3. Create Audience Personas

Findings

Interview highlights:
  • Most individuals understand that they are not exposed to all content in their social media feeds, but they do not understand why. 
  • People do not want to be exposed to more in-group content and are frustrated by what the algorithm shows them. 

Defining the Problem

Our primary audience (educated individuals who are exposed to social media algorithms) is largely unaware of how social media algorithms quantify their behaviors and unaware of what actions they can take to mitigate it. 




UX Design 

Project Goals

  • Explain about how social media algorithms work
  • Explain how user interactions have an impact on social media content
  • Get users to reflect on their automatic online behaviors 

Product Features

  • Survey to incite personal reflection of online behaviors
  • Interactive background on the research experiment to create experiential learning
  • Link to additional external resources for further education

Information Flow/Storyboard

  1. Social media ranks posts based on our behaviors. 
  2. We often behave automatically online. Our automatic behaviors reflect our subconscious in-group biases. 
  3. We ran an experiment on movie recommendations to test this. 
  4. The way we act doesn’t always reflect what we want. 
  5. Algorithms create a double penalty on out group content
  6. What we see in our feeds does not reflect the world as it really is

Early Prototypes

             



UI Design 

Fonts and Color Palette


Visual Elements

  • We chose the metaphor of reflections and watching to allude to the algorithm observing users’ micro-behaviors. 
  • We used illustrations of digital elements and networks to be consistent with imagery throughout CAAI work. 

Graphics






Usability Testing and Solution

Iteration

In our design iterations, we made adjustments to simplify data graphics and make our user interactions more consistent throughout the page. 



Final Product

The site design can be accessed here. 



Project Takeaways

Lessons Learned

  • Motion and movement within the page should be intentional
  • Gestalt principles, such as  contrast, for drawing emphasis and attention is crucial
  • We should leverage more structure when designing usability testing


Challenges

  • Timeline management: Overall, our design process took more iterations than necessary, and with a more structured process and design system, we could complete the project more efficiently.