Master Thesis
Algorithm Fairness Label
A concept how design can be used to deal with algorithmic bias.
Research
How can design be used to minimize algorithmic biases?
Approach & Visualization
- Designing an understandable representation of the problem to make the debate about algorithmic systems accessible to the general public.
- Five communication guidelines to provide guidance in dealing with algorithmic biases.
- A concept of a seal of approval to expose algorithms that are operating unethically.
My responsibility: Concept & Design
Hochschule München | Master Advanced Design
2018
Digital ethics research
to deal with algorithmic bias
Problem
- Subjectivity of algorithms and the influence of their decisions
- False, discriminating results
-> Algorithmic Bias
Solution
- Communication Guidelines for AI and Algorithmic Bias
- Algorithmic Fairness Label
The label
Algorithm Fairness Label –
A kind of seal of approval for automated systems with the aim of detecting unethical algorithms in terms of fairness and transparent communication.
Prototype
- Evaluation according to the degree of bias of the algorithm.
- Measured by “reverse engineering” (analysis of the result and reverse engineering of the algorithm).
Communication Guidelines
for the evaluation?
Sensitivity
To make an algorithm fair, it must be clearly communicated that you are dealing with automated decision making processes.
Transparency
Informations have to be retraceable and communicated transparently.
Accessibility
Information has to be fully accessible. Unpleasant information has to be communicated equally.
Human Judgement
Humans have to be integrated into the decision making process of algorithms.
Ethical Importance
Potential consequences caused by the algorithm must be assessed.
Infographic
With playful animation, simplified shapes and friendly colors, the process of algorithmic decision making is presented here. The algorithm runs through various parameters, such as gender, age, etc. to make the decision.
Exhibition
2019
