The problem is that their models are so broken, they require an insatiable amount of invasive training data to try to correct for their erroneous conclusions. T

The problem is that their models are so broken, they require an insatiable amount of invasive training data to try to correct for their erroneous conclusions. Their failures drive even more affronts to privacy. The cycle continues. Then they sell us household gadgets to sink their hooks deeper into our lives. Microphones listening to conversations won't be enough; soon we'll herald the coming of new "smart" appliances like the Google Toilet, which plays your favorite music, recommends better posture, sprays you off when done and stealthily analyzes your stool samples for markers to see if the Captain Crunch ads they bombarded you with over the last week led to a conversion. It doesn't work, but failure is clearly no impediment to progress.