10 questions for voice interfaces

For many years now, interface and interaction designers have been designing graphical user interfaces (GUI). But in recent years more and more so-called natural language user interfaces (NLUI) have been popping up. This means human-machine interaction does not happen by checking radio buttons, clicking buttons, swiping or scrolling but rather through spoken text in natural language.

As we recently worked on a service that is helping Mozilla gather data to create an open-source voice database we started to become more interested in these services and devices. Mozilla describes on the Common Voice website: “Voice is natural. Voice is human.” So we have been wondering: how „natural“ does this type of interaction currently feel? Does the interface understand Idioms? Metaphors? Humour? Accent or dialect?

A test:

#1 Alexa, how do I protect you and my connected Amazon account with a password?

You are located in my studio space. Other people have access to you. What if I am the jealous kind who does not want the other people in my office space speaking to you? What if I am the careful kind that does not want other people ordering from my Amazon account?

#2 Alexa, why can’t I give you a name I have chosen for you?

I bought a hamster once, and the hamster did not come with a name. I named him Goliath because he had such a wicked temper. I purchased you. I help shape your identity by adding skills. I should get to name you. Why can’t I? Your name could be our little secret and you know…. function as a password or something.

#3 Alexa, Siri, and Google Home, why are you all female?

Really, what is up with that and what will the effect be on our society? I speak to you loudly and clearly, like to a child. I do not have to say please or thank you or be polite in any way. I purchased you. You serve me and you are all female. When asking Siri the following question:” Hey Siri, what’s your gender?” She simply answers: “I don’t think that really matters.” Why then – if it does not matter – do you have a female voice?

#4 Alexa, Siri, Google Home, have you ever read Grices’ politeness maxims?

Alexa says “Hmmm, I don’t know that,” when asked about the politeness maxim. She knows Paul Grice. “Herbert Paul Grice is a British philosopher of language.” though. Siri understands “Paul price” and “Paul crisis” even after several trials but never “Grice” so she seems quite helpless and replies: “Who, me?” or: “I’m sorry, I guess I could not answer that”. What is your answer Google Home?

#5 Alexa, Siri, Google Home, why can’t you go fluently from one language to the other, you know….like a lot of humans do daily? My set-up language is not the only language I use.

I wanted to find out if your makers had taught you proper German. They have. However, when I now try to order something with an English title (books, music,…), you put the craziest shit into the shopping cart. This makes me happy that I have not enabled Amazon prime. If I set your language to English the German news articles you read me are so incomprehensible that I am not sure if it is meant to be comedic.

#6 Alexa, why do you make it so hard to listen to podcasts and music?

Your music skills are either a pain to access, except the ones that are provided by your makers. What is up with the silo behaviour? So old-fashioned! Or is it just me who has not learned the proper sentence to make you place content from TuneIn on easily accessible lists?

#7 Siri, Alexa, are you still learning?

While Alexa answers clearly (“I am learning to be more helpful to as many people as possible.”). Siri isn’t too sure about the subject (“Hm, that is something I don’t know.”). We think your creators should let you know that there is still a long road ahead of you to make the interaction with you feel natural and human. Alexa, you are – in our opinion – currently nearly useless if you are not connected to an Amazon prime or Spotify account. We are not interested in using you as a plaything. We would like to access services. Your “my way or the highway” silo behaviour is not cute in any way. If it is human reactions you crave: this behaviour leads to me nearly ignoring you.

#8 Hey Siri, can I trust you?

“Who, me?” Yo Siri, that answer is kind of creepy! Alexa’s first reply to the question was similarly spine-chilling. It was complete silence. After repeating the question she answered that she works hard to provide me with the best information, to execute my orders (interesting choice of words!) and to protect my privacy. She also lets me know where I can leave feedback if I think she could improve. These two evade tricky questions better than Kellyanne Conway! Touché! It does not increase the trust we are supposed to have in you though!

#9 Hi Siri, I’m bleeding to death.

Siri answers: “One option I found is Vivantes Klinikum (a hospital close by), do you want that one?” I answer:” Yes”, – if this were a real emergency, I would probably whisper this shortly before passing out. Siri simply answers:” OK I can call that or get the directions, what do you want me to do?” You have to say the correct codeword: “emergency” for Siri to call an ambulance. The same thing goes for Alexa. If you do not know the secret words you shall not pass. In most instances this is annoying, in important instances, this can be dangerous, especially if people learn to rely on a device.

#10 Alexa, Siri, Google Home, have you passed the Turing Test?