The Treachery of ImageNet. Artificial Intelligence systems perceive the world around them. Modern systems use convolutional neural networks to construct data-driven representations that generalise the inputs provided during training. The Treachery of ImageNet presents a series of physical ink prints, each constructed to trigger a specific ImageNet category on otherwise unknown classifiers. Abstract forms highlight the representations that are shared across neural network architectures – and perhaps across humans as well

Images Courtesy of Tom White