"Smart Replies" act human so you don't have to.

One of the features demonstrated in Google's upcoming Allo messaging app is Smart Replies: AI will analyze the conversation and context, and suggest some possible ready-to-send responses. The image above shows an example, where "aww so cute" and "Love the daisy!" are suggested by the app based on what it parses from the received photo. This is certainly technologically impressive, where the AI is doing a fine job impersonating a human; the problem is that I don't want to outsource my humanity. I want my what I say to come from me, not the suggestions of a bot - and I want to know the same is true of what is said to me by other flesh-and-blood humans, even if it's via electronic means. Autocorrect features to fix typos or speed up text entry are fine, as they don't interfere with human-originated intentions - even offering quick-responses like "running late" or "busy now" is handy for quick logistical messages. But when AI strays into how I should emotionally communicate with others, that's too far. Engadget's Nicole Lee agrees: don't send me Smart Replies!

No comments: