2021 - 22
For early career professionals from non-traditional backgrounds, a recruitment tool in the form of a digital game. Diverging from conventional solutions, it places an emphasis on capabilities rather than keywords and arbitrary attributes. BalanceAI is designed to mitigate algorithmic biases within the assessment process.
| Tags | Interface Design, UX Design, Interaction Design, System Design |
|---|---|
| Research methods | Ethnography, Domain Expert Interviews, A/B Testing, User Interviews |
| Primary tools | Unreal Engine, Adobe Suite, Blender and Python |
| Industry collaborator | Rainbird Technologies |
Games and machine learning can replicate real-world scenarios. A user's aptitude can be measured based on their interactions with such simulations.
AI can be used to generate the signal / noise ratio of a resume from observing a users gameplay (which contains a higher amount of contextually relevant information).
In conversations with seven recruiters across diverse industries, it became evident that automated systems played a pivotal role in early career hiring due to resource constraints. A combination of applicant tracking systems (ATS) and various AI tools were commonly employed.
By applying for jobs and interviewing applicants, it became apparent that these tools favoured specific physical attributes, educational backgrounds, and keywords within applications. These were incorrect criteria to optimise for, particularly for disadvantaged individuals from diverse demographics.
Further investigation into these AI tools revealed that the datasets used for training mirrored past biases and societal norms. Additionally, these datasets suffered from class imbalances, perpetuating systemic biases and resulting in oppression via algorithms.
Diagram summarising the research stage