In a move being hailed by as “clever” by some and “creepy” by others, Amazon is moving to make the AI that powers Alexa just a little more sensitive to consumers’ needs.
And, as it turns out, a little more empathic, or at least apologetic, as well.
According to a source familiar with Alexa, or the Amazon Echo, AI researchers are developing better natural-language-processing capacity such that Alexa won’t just understand the words customers are using – “she” will also hopefully also be able to learn to read the tone that lies behind them as well.
The theory goes as follows: Alexa is designed to learn more about the consumer the more the consumer uses the product. A user who lives in Boston will deal with an Alexa who knows that fact when he asks about how the Red Sox are doing. A user whose favorite thing is listening to Taylor Swift will teach Alexa about her preferences about music in general.
But Amazon recognizes that AI makes mistakes, and that more mistakes will be made early on. And normal humans don’t take being denied by AI well, as evidenced by every person that ever yelled at their car for failing to connect a call correctly.
Alexa’s marketplace is getting increasingly common, and Google’s Echo competitor is already generating a lot of buzz for being that much smarter and better with human being’s voice commands. Despite the fact that said competitor is mostly a rumor at this point, if Echo can’t do better, well, at least it can apologize for when it is worse.