Is AI biased?
-Installation about Artificial Intelligence
Bias in machine learning occurs when the results of the research cannot be generalized widely. It results from how the data is obtained and the choices that are made while training the data. In this project, we wanted to explore the correlation between input and output data and try to understand how one can take control of the final result by training their own data set. We collected and built our own data set to highlight the ethical issues and implications of the use of Artificial Intelligence.
*Collaboration project
-My role as a concept creator and ML technologist
Process
55 diverse students, faculty members, and staff were photographed at the Pratt Institute.
The curated data was trained in Runway.ML. After about 4 hours, the trained model generated a new set of 50 images.
Overlayed the images generated by the model into 4 sets of images.
Installation
To replicate the aesthetics of a convolutional neural network, we decided to print the generated output in a sheer fabric (organza) and hang them layer by layer.