Having an opinion makes you more interesting, even as an assistant…This is the power of machine learning. One of the most interesting things we’re going at is how do you design an assistant that feels like you’re having a conversation with someone
Opinions About Self Versus Others
This isn’t news per se. Alexa has been offering insights into her preferences from the beginning. So has Siri for that matter. However, the types of opinions have expanded beyond Alexa’s scripted nature. Perez points out that Alexa is expressing opinions about other things. Where do these opinions originate? Is it machine learning or programmed?
Perez clarifies that there is a difference between facts, recommendations and opinions. Most people ask Alexa or Google Assistant for facts. But, there are also the many instances when they treat Alexa more like a human being. For example, check out YouTube to see people asking Alexa what she thinks about Siri, (answer: “I like all AIs”), to name her favorite color (answer: “infrared is super pretty”), and her favorite time of day (answer: “definitely 11:07 am”).
Asking about Alexa’s favorite movie first yielded Edward Scissorhands and then later it was Princess Bride. Both were opportunities for Alexa to make pun-based joke. A consistent response came when asking about her favorite smartphone, “definitely Galaxy S7 Edge.” That is particularly interesting given that Samsung Galaxy phones include both Google Assistant and Bixby, but do not support Alexa. The video below shows a few of those query responses on on an Amazon Echo Show. I was able to recreate a similar experience on a standard Echo unit.
Opinions May Be Transitory
Interestingly, the opinions sometimes change. I was able to recreate Perez’s example about Alexa’s favorite beer (answer: definitely Budweiser), but as you can see in the video, she eventually switched to Bud Light. The people of St. Louis must be thrilled. Similarly, the first time I asked about her favorite soda the answer was, “I’m glad you asked. It’s all about Dr. Pepper.” After that, the answer was consistently Coca-Cola for several attempts before returning to Dr. Pepper and then back to Coca-cola.
So, why do opinions change within a few minutes? Who decides on the ultimate nature of these opinions? How will these opinions impact consumer behavior if at all? These are important questions for users and for brands that may benefit from a favorable mention.
How Opinions Differ from Recommendations
There are different types of opinions. The implications of expressing a favorite color differ from favoring a specific television show or branded product. Many people just assume that consumers will ask their voice assistants for factual information. That is just one type of response. Voice assistants can provide more value to users by making recommendations about products and entertainment. Assuming the voice assistant is trusted, it will save time for consumers. That could be viewed as even more valuable than executing simple tasks.
Opinions are something different. In one regard, recommendations are opinions. However, opinions don’t have to involve a recommendation. Just because Alexa likes Bud Light doesn’t mean that is her recommendation for you. In fact, when you ask Alexa to recommend a beer or soda, she apologizes and doesn’t have a recommendation. Alexa’s opinions are simply a part of her nature or what was programmed to be her nature. It may be viewed as product placement. The opinion may raise awareness. However, it doesn’t rise to the level of social proof. Recommendations are more important.
How Do Recommendations Change the Nature of Assistants
This brings us back to Amazon’s Marc Whitten. Both TechCrunch and Variety report that Whitten is contemplating how Alexa through Fire TV can offer value similar to what Blockbuster store clerks used to do for patrons. This should surprise no one that has ever shopped on Amazon.com. It has a “your recommendations” section which is tailored to your shopping profile. Neflix recommends movies based on your past viewing history.
Recommendations can be based on what the algorithm believes you will like or they can be sold as sponsorships. That is the option of the platform. I am not saying that Amazon is doing this or has plans to. I am pointing out that Amazon and Google have the option to do it. And, offering recommendations unpaid or paid may actually add value for the user. Beyond recommendations, voice assistants may instead be empowered to make choices on behalf of consumers.
This is precisely why Amazon, Google, Apple, Microsoft and Facebook all covet the opportunity to become a consumer’s voice assistant of choice. They all seek to have direct influence over the consumer experience, their choices and what recommendations they receive.