> I then had the pleasure of experiencing some of these much-lauded “human touch points” first-hand on my travels back to NYC from abroad > the clerk, who had

> I then had the pleasure of experiencing some of these much-lauded “human touch points” first-hand on my travels back to NYC from abroad > the clerk, who had apparently been craving some human touch points herself, had left her post to converse with the other clerk on the far side of the store. I ventured across the store to give them my money and was made to feel like I had barged into their private living room, demanding to be served Was "abroad" France? Sounds like it (I'm French). Anyway, the problem with AI isn't that we'll miss the so-called, and probably non-existent, "human touch". It's that it doesn't work. (Examples given in the article, of vending machines and automated checkout, don't have much to do with AI). See translation for example. It's not that automatic translations are bad, it's that they're incomprehensible: you couldn't guess what the text means if your life depended on it. In Europe, Amazon offers to translate your listing in all marketplaces' languages (English, German, French, Spanish, Italian) when you create it for just one; that produces gibberish and results in zero sales, until you use an professional (human) translator. Google translate, despite their recent bragging about a revolutionary update, is no better. Small example: in French, "coup de coeur" means something you love at first sight ("j'ai eu un coup de coeur pour cet appartement" => I fell in love with this place at first sight). But Google translate translates "coup de coeur" with "heart stroke" https://translate.google.fr/#fr/en/coup%20de%20coeur https://translate.google.fr/#fr/en/coup%20de%20coeur which is beyond wrong (in no context ever can this meaning ever occur). I noticed this because on some forum a user was describing his blogging activity as writing about his heart strokes and I wondered how many he had had and how it was possible to still blog about them. Back to human interactions: machines are great when the outcome is discrete and completely predictable; using a vending machine, buying things from Amazon, checking in at the airport, etc. But when some creativity is needed, machines can become super annoying; calling customer support and finding oneself interacting with a robot makes you want to murder someone (but there isn't anyone left to murder).