Recently, Tech Crunch reported that Alexa, Amazon’s personal assistant, is about to start having her own opinions about things. They said that in an effort to make her more human-like she needed to have opinions that she comes up with on her own through what’s called machine learning. Not that she doesn’t already give opinions, but right now they have been programmed in by humans. In the near future, Alexa will start offering her own thoughts on things like TV shows, movies, books and more. But how do we know why she picks the things she recommends, now or in the future?
A recent video showing someone asking Alexa what her favorite beer is, with her answering “Budweiser,” got us thinking. Why did Alexa fall in love with Budweiser? It can’t be because it’s the best selling beer (because it’s not, that’s Bud Lite and Alexa stated as much when I asked this specific question). To see just how much Alexa loves Budweiser, I put my own Alexa to the test. I asked not only what her favorite beer was, but also “what’s the best beer?”
It turns out that not only is Budweiser Alexa’s favorite beer, but it’s also the best beer in her opinion. The more I asked the question, the more she doubled down and stuck with her pick, showing a wide range of responses pertaining to Budweiser.
While I did have a lot of fun peppering Alexa with the same question over and over again, what is concerning is the level of influence Alexa has and will have over people’s choices now and even more so in the future. Furthermore, Amazon loves monetizing things, and who’s to say that they wouldn’t start selling Alexa’s opinions (or haven’t already) to the highest bidder so that she can name drop their product over and over again. Only large companies would be able to pay for this sweet product placement anyway, and maybe they already have.
So the next time you ask Alexa for her opinion, take it with a giant grain of salt. While Alexa is still just a machine, she can easily be influenced and that influence might be due to someone’s marketing dollars rather than a computer’s machine-learned opinion.