top of page
Abstract Lines

AI Bias Detection

This scenario-based eLearning experience equips social media community managers with data analysis skills to critically evaluate and respond to signs of social bias in the platform’s AI-based recommendation engine.

Audience - Social Media Community Managers

Responsibilities - Instructional Design, Visual Design, eLearning Development, Action Mapping, Storyboarding, Prototyping

Tools used - Articulate Storyline 360, Lucidchart, Figma, Google Gemini, ElevenLabs

AI Bias Detection Final Project Mockup

The Performance Need

CrossRoads is a fictitious social media company using state-of-the-art AI technology to provide adaptively personalized content to its users. Due to the influence of social bias on user preference data, CrossRoads’ recommendation engine was exacerbating the harmful effects of social bias (e.g., discrimination, echo chambers, etc.).

Although community managers overseeing the engine’s recommendations had the basic data analysis skills to promote overall engagement, they were unable to distinguish signs of social bias from innocuous user preferences. I identified a performance need for community managers to identify and respond to signs of social bias in the platform's recommendation engine.

The Learning Solution

My solution was a scenario-based eLearning experience allowing learners to:

  • interpret multifaceted performance metrics,

  • strategize in response to these metrics by weighing competing company interests,

  • explore the consequences of their decisions in a low-risk environment.


Action Mapping

In lieu of a subject matter expert (SME), I independently researched the topic of AI bias and social media recommendation engines. I utilized Google Gemini to act as a mock stakeholder, helping me to identify an actionable business goal pertaining to AI bias for the company.

I then created an action map to derive concrete performance tasks from this business goal. This results-driven analysis allowed me to and strip away unnecessary information and pinpoint key learning goals to effectively solve the performance need. 

AI Bias Action Map

Visual Storyboard and Interactive Prototype

From here, I developed a visual storyboard, drafting the on-screen text and programming notes while experimenting with visual layouts using Figma. ​I then used Articulate Storyline to prototype the module's interactive elements.

AI Bias Storyboard
AI Bias visual mockup
Design Features

Branching structure - Using the action map, I organized the module into a branching structure, prompting learners to make authentic performance decisions with realistic consequences. As learners move through the module, they must make strategic decisions based on limited data, unveiling pertinent information depending on their choices. To achieve continuity, I used triggers, variables, and conditions.

Digestible look and feel - I opted for a clean visual format to reduce learners' cognitive load while deciphering ambiguous information (as part of the performance goal). I used simple icons to help users visualize their findings and to maintain visual interest without distracting from the information-oriented task. 

Interactive learning experience - I incorporated a high level of interactivity using:

  • hover and click interactions to progressively reveal information on each slide, and

  • a tab interaction to summarize the module's key findings.

These interactions serve to retain learner's attention and segment information into chunks that learners can digest at their own pace. 

Screenshot of Articulate Storyline triggers
AI Bias Animation Prototype


From the interactive prototype, I fully developed the module in Articulate Storyline

For the introductory video, I sourced stock video clips from Pexels and used ElevenLabs AI text-to-speech generator to create the voiceover. 

bottom of page