this google+ thread of ml researchers discussing deep learning in 2012 (when it became a “thing”) is super interesting. lots of folks discussing bias, not in terms of social issues but in search of the “perfect dataset” best mirroring real world variance https://t.co/dB4oVOkqxu