The Multi-Dancing Boxes is my research project for the MA Digital Direction program at the Royal College of Art. The project is affiliated with the topic of AI ethics, bias, and human-machine relationship. I explore how the diversity of human culture could influence AI vision by experimenting with the initial stage of interaction between humans and AI.
On the methodology side, I trained the dataset by selecting a group of people with different backgrounds and asked them to describe the objects that they saw in my video footage. I then collected all the data based on their input to create 4 or 5 machine-learning models that represent each culture and display the combined outcome as the Multi-Dancing Boxes. The object recognition capability helps to capture human’s cultural bias during the first human-machine interaction.
Through this exploration, the project questions how we as a human with different cultural perspectives, describe and categorize the world surrounding us from unfamiliar objects, sounds, situations, to the unknown. The range of differences we observed shows how important inclusivity is, especially in the development of new areas such AI.
On the methodology side, I trained the dataset by selecting a group of people with different backgrounds and asked them to describe the objects that they saw in my video footage. I then collected all the data based on their input to create 4 or 5 machine-learning models that represent each culture and display the combined outcome as the Multi-Dancing Boxes. The object recognition capability helps to capture human’s cultural bias during the first human-machine interaction.
Through this exploration, the project questions how we as a human with different cultural perspectives, describe and categorize the world surrounding us from unfamiliar objects, sounds, situations, to the unknown. The range of differences we observed shows how important inclusivity is, especially in the development of new areas such AI.
