Assignment 8: Alexa, Talk To Me

This was one of my favorite assignments. Learning how Alexa works behind the scenes was very interesting. Where this technology really interests me is in becoming a personal assistant to anyone. That is the focus I took in writing my review.

Vocal User Interface: The New Personal Assistant

Imagine a future where all aspects of our lives are controlled by a virtual personal assistant (VPA). They wake you up in the morning, they get coffee going, they give you the day’s to-do items, tell you what the commute is like and even drive you to work. Should anything in your day change, you simply tell your VPA what has changed and they will adjust accordingly. They can do your shopping, pick up the kids and even help you do your taxes. This is a future that I am very excited for. As someone who lives somewhat sporadically I would love a VPA to help run and organize my life. We are quite a way from this future though, as voice interface is still a new technology. Even so, voice interface has already started to embed itself into our lives. “Gartner predicts that, by 2019, 20 percent of all user interactions with the smartphone will take place via virtual personal assistants(VPAs).”

Current State of Voice Interfaces

While my interests lie within the realm of a VPA at home it seems that some of the biggest advances in VPAs are within businesses. Luxury retailer, Liberty is already using a VPA in their warehouse. “Staff are issued with headsets and a microphone. A voice-controlled application from Voiteq interfaces with the firm’s enterprise resource planning system. The voice-controlled app issues commands to a central server about which products should be picked. The system also works on a two-way basis, issuing confirmation about which goods have been selected.”

It will be through business innovations that the VPA will make big jumps into the home. Major companies are already starting to integrate Amazon’s Alexa into their software. “Ford, Whirlpool, Lenovo, GE, and Dish are a few of the brands announcing plans to add Amazon’s voice-activated helper to their hardware.” As these major companies roll out these changes Alexa will begin to be able to do things like monitor the laundry or start your car. As the technologies advance Alexa, and other VPAs will begin to interface with all aspects of our lives.

Despite major advances the technology that VPAs require, the technology is still in its infancy. “Technologies upon which virtual aides rely, like speech recognition and deep neural networks, have improved dramatically over the last several years. That has translated into big improvements in software like Alexa and Siri, according to Carnegie Mellon’s Black. But they still lack a thorough understanding of speech, which Black expects to be the next area of focus for companies like Amazon.” While these technologies are leaping forward as companies pour resources into them, many industry professionals think it will be some time before we see fully integrated VPAs. There are a variety of challenges before we get there, not all of them technological.

The Challenges

There are several key challenges to voice interface and virtual personal assistants become integrated into the average person’s day. The first is the ability of the machine to understand speech. Right now, only one command can be given at a time and those commands are limited to the utterances that are programmed in. Take for example these more complex commands,

• “Deliver the items on my weekly shopping list Friday afternoon before six and check for lowest prices on the produce.”
• “Find the photos from last Saturday’s picnic, adjust for white balance and send me only the shots with me and my brother for review then email them to my mother.”

These commands are ones that any personal assistant might get but they require a more nuanced understanding of language. Another major challenge in the technology is being able to determine which app, or skill in the case of Alexa, to use. Currently, we must tell Alexa which Skill to open but to be able to handle more advanced commands like the ones above Alexa would have to be able to access multiple skills and make the determination of which at access. Basically, Alexa and other current VPAs need to learn how to use Apps like a person. The last major challenge to the future of VPA’s is a business/ethics one. Right now, the three major VPAs are Amazon, who is leading the way, Google, and Apple. As they all develop their own systems some industry professionals, like Peter Markey is former CMO of Aviva, think “There will probably be a VHS or Betamax moment where only certain voice-activated systems survive.” As these companies compete for our loyalty they could use their VPAs to gather data about us. So, is that VPA there to help us or to help the company? Can we trust our VPA to consider all options and choose the best for us? Or will it choose from a set of options that help the company? As the technology challenges are overcome, and they will be at some point, these are questions that we will have to answer and figure out. Regardless of how these issues are handled, everyone agrees that VPAs are in the future.

Articles Reviewed:

• Siri, Cortana, Alexa and Google Assistant are just the beginning: Voice is the future
• Amazon Is Already Winning the Next Big Arms Race in Tech
• The future will be voice-operated but digital assistants need to learn apps
• Gartner Says by 2019, 20 Percent of User Interactions With Smartphones Will Take Place via VPAs

Part 2: Create Alexa Skill

Part two was tons of fun. In addition to the utterances that I needed for the assignment I had fun making a bunch more from stuff about me to just goofy things.

Everything worked great when I tested it.

Alexa, Talk To Me from Jim Murphy on Vimeo.

My favorite part about it all is Alexa saying my name as a question. That’s because in class we were seeing if Alexa would read different punctuation and we were testing it on the code we all used to create our skill. So now Alexa isn’t sure who I am.

'JimMurphyIntent': function() { this.response.speak('Jim?'); this.emit(':responseReady'); },

I would be very curious to know how we can modify that code to do different things. There is so much opportunity in learning how to write functions and intents that I can then build into skills.