The Politics of Classification – Information Architecture and Artificial Intelligence
Access ImageNet, created in 2009, 14 million images and 21k categories strong, and you’ll find yourself staring at a curious top-level taxonomic structure that orders the world into plants, geological formations, natural objects, sports, artifacts, fungi, persons, animals, and “miscellaneous”. Or take a look at one of the other datasets used for computer vision, such as IBM’s Diversity in Faces, Humans in 3D, or the Flickr Diverse Faces Dataset.
These datasets are the foundations used to build the AI models that give us self-driving cars, delivery by drone, or face recognition on our phones. But also misdiagnoses, discrimination, racism.
Tech-driven narratives assume AI to be a computationally solvable problem: we just need to fix the model. Either the algorithm can be tweaked, or more data is needed, or both.
Information architects know better: this is a wicked problem, this is political, the world resists being reduced to mathematical smoothness.
Join me to discuss the structurative, agentive, and evaluative nature of AI and the role that information architecture and experience design should play in making better use of AI, and AI better, by considering three major questions:
Now that AI can do <x>, what will it be like to use it to do <x>?
The individual experience of a world in which AI exists
What are the consequences of AI being able to do <x>?
The vast externalities descending from the existence and use of AI
What politics does this AI bring into the world?
The even larger implications of the non-neutrality of AI