Computer vision has made great strides lately; human-level face and object recognition can run on a laptop's GPU in real time. Segmenting a rendered webpage, cl

Computer vision has made great strides lately; human-level face and object recognition can run on a laptop's GPU in real time. Segmenting a rendered webpage, classifying regions as ad- or non-ad, and overwriting (in a way invisible to your canvas renderer the former, by comparison, is a walk in the park. The client will always win. Browsers are user agents, not server agents. https://www.youtube.com/watch?v=1C-2dKe9NjI https://www.youtube.com/watch?v=1C-2dKe9NjI