Shhhhhh, she’s listening

Published 3:00 am Monday, May 21, 2018

By Greg Price

There are many questions about artificial intelligence. Among the questions are various approaches to assessing whether a manmade device is able to reveal intelligent behavior, the Turing Test is a popular approach. Software, well-designed and properly-scripted, can mimic human responses rather easily nowadays. In fact, the notion of impersonating a human via software is so common that we are often unaware that we interact with “smart devices” daily. From where does this artificial intelligence arise? Does the program, the device need to be self-aware, or, does it simply have to be so well-designed that it fools most humans? Do we need another test?

Google’s Duplex and Dialogflow answer questions in natural language, makes voice calls on behalf of their users, converse and interact with humans with ease. Last week, Google Duplex was exhibited and the software made calls to humans and scheduled restaurant reservations and a hair appointment, all with an incredibly realistic voice.

Sign up for our daily email newsletter

Get the latest news sent to your inbox

Are these tools artificial intelligences, or, just very clever tools?

Well, I’ve made an observation. In fact, I’ll refer to it as the pronoun test. When my friends, family, or even myself, start referring to ethereal software-generated voices as he or she, then perhaps we have some rendition of a manmade intelligence. Whether you’re talking to Siri, Google or Alexa, how do you refer to “it” when actions are not performed properly? Did she or he provide the wrong directions, or, did it not work correctly?

Ponder that for a while. I found myself in an argument recently with both my truck and my phone. At one point, during said argument, my watch interrupted and decided to take sides against me as well. I’m confident that I looked like a madman from a distance. As I argued feverishly with “them”, I arrived home.

Exiting the truck, one personal digital assistant went away as I turned off the vehicle. However, the hotly-debated issue with my digital devices escalated into a major concern: my house, was, in fact, “not responding”. It seems the truck, the phone, and, the watch were all correct, my house was “down”.

I tried to contact some of the home automation equipment via my phone. No luck. I tried to connect to the house via the truck, because, well, why not? Nothing happening. The watch, apparently feeling neglected, tried to connect to the house as well, offering, a third confirmation, “the house is not responding”.

Standing outside my house, I realized that I was truly entering interesting territory, my house had “rebooted”.

The home automation has a fail-safe measure: lose power and internet connectivity and it enters panic mode. Due to construction, the house was temporarily without power and internet access, the backup batteries were depleted. The house, too, was arguing with me, it was in panic mode and I couldn’t disable the locks and alarm system.

So, the phone, the watch, the truck, and Siri were correct, my house was “not responding”. Who would have thought my house would have disappeared from the grid.

So, what does this have to do with artificial intelligence? I had to call my wife and ask her how to get into the house. Apparently the keys still work. She was correct.

Once the house rebooted, Alexa and I had a serious conversation about notifications. I’m certain she was listening, but, did she understand?