Why We Need to Think About Assistant Neutrality

Charles Jolley
Oct 5, 2016 · 2 min read

An edited version of this appeared on VentureBeat.

With the release of Allo last week (and how Google Home today), we’ve officially entered the era of the assistant. Every company that owns a major tech platform is betting that assistants will be an important interface in the post-app world. With more and more assistants coming alive it becomes more and more important to consider the source of the knowledge assistants have to share with people.

While most assistants today are mostly used to play music or entertain our kids, the next generation of assistants will go far beyond. They will provide us with recommendations, guidance, and even make decisions for us. Google even says they will replace search.

And yet, the assistants offered by major companies today take a very limited approach to their data. They often pick one winner per category of answer, preferring their own properties when possible and non-competitive alternatives otherwise.

Google’s Assistant is the newest and probably the most advanced, so let’s use it as an example. Here I am trying to plan a date:

Using the Google Assistant to plan a pizza date

Notice how every answer you get here leads you to a Google property. You get Places, the Knowledge Graph, or one result from Google Search (the ultimate I’m feeling lucky!) The assistant could have used Yelp, TripAdvisor, Michelin, and others to provide better answers to my questions, but they are no where to be found. There are no third parties directly informing the Google assistant.

(I should note that Google is promising an SDK. Similar to Amazon’s Alexa, you will have to ask for third party services by name; their data won’t inform the assistant’s results. Siri is taking a slightly different but similarly restricted approach.)

This problem isn’t limited to Google. Alexa was built to encourage people to buy more on Amazon. Apple is bought Siri to get you to buy more hardware. Virtually every assistant built by a major company exists primarily to serve some other business need, leading them to limit the data they use to non-competitive sources.

The era of intelligent assistants is upon us. That is why this is so important to think about how assistants will determine which sources to consult. How will we know we’re getting the best answer possible when our assistants recommend TV shows, flights, hospitals, medical treatments, or jobs? We need to start talking about assistant neutrality now.

Humans make the best decisions when they can take into account many different points of view. We need our assistants to do the same.