curlicuecal:

Now I’m reading the google labs article on how they developed the Gmail Smart Reply feature (which suggests replies to incoming emails) and it’s fascinating–

all the little hurdles they had to overcome– like getting the neural network to hone in on the most useful fragments of the incoming message, and teaching it to identify semantically different responses instead of generating, say, three variations of “let’s meet tomorrow” to reply with

ANYWAY:

Another bizarre feature of our early prototype was its propensity to respond with “I love you” to seemingly anything. As adorable as this sounds, it wasn’t really what we were hoping for. 

HELP, THE ROBOT IS CUTE.