“Hey Alexa, What’s Up?”: Studies of In-Home Conversational Agent Usage

The researchers focus on just four main subjects surrounding VAs as the technology is still in its early days of adoption and means of data streams available are still limited. 

(Focus is also limited strictly to AMZN Alexa Devices)

  • The rooms the devices reside in
  • The daily routines asked of Alexa , Task Completed or Failed to identify / Complete
  • Acclimating to Alexa post purchase and advantaged of an expanded ecosystem of devices
  • Current limitations with Human-Device interaction , specific focus on children working with device

Devices have been shown to work sufficiently better when integrated in an ecosystem of other devices so that full coverage of the home is available for Alexa to pick up commands

Many people who own an Alexa end up having more than one spread throughout there house, 42 percentage of Alexa owners end up owning more than one device (Kernel Density).

Most common Location for VA device Placement –

Kitchen, Bedroom, Living-room (Bedrooms being the most popular location)

Usage peaked early in the morning [9:00am] and in the afternoons, not Surprising ( Weather in the mornings, Timers in the night )

Instead of looking for phone when the impulse came to mind, many VA device owners noticed that they instead would just call out a voice command to Alexa asking to relay the time back to them.

Parents derived joy from watching their children interact with Alexa, but children have difficulty interacting with the device usually lacking the basic skill to properly format or pronounce there questions / commands.

A specific interaction referred to an example when participants children asked the device to spell out a word for them . After Alexa immediately prompted a reply answering the child’s question the parents expressed joy and trust in watching the device help in the learning process /// ?.?

Va devices as personal tutors in the future ?

Feedback from the research participants was overwhelmingly positive, because of the small physical profile of the devices many of the participants guests never were never even aware of the Alexa device in the room.

[  “Only one household, mentioned a negative experience from introducing Alexa. A visiting family member requested that the device be unplugged because the relative was concerned that the device was recording their conversations.” ]

Improvement Suggestions :

Data Mining to offer new features ; Collect a larger amount of meta data on the specific use cases used most frequently by the device owner . This can create the opportunity to help Alexa’s algorithm improve its features and suggestions to personalize the experience that best fits the user it is trying to serve.

Leverage knowledge of Place ; By giving Alexa context of the location it is being placed to operate it can adjust and modify to work best in a certain type of environment. Depending on the room the device is placed in such as a Bedroom Vs. a Living Room , particular commands can be predetermined to be most likely asked producing better quality suggestions for new uses.

Integrate connected devices ; While the vast majority of VA Device owners also own smart phones , little to no options are available on the market for integration between both the devices. This aspect can open up the opportunity for many improvements in the overall experience and results the framework .

Improve the quality of Child and device interactions , This may take time and advances in the NLU / NLP Algorithms throw increased diverse interactions with Customers.

Voice Commerce: Understanding Shopping – Related Voice Assistants and their Effects on Brands

[ “This study sheds light on the  potential  impact  that  the  diffusion  of  shopping-related  voice  assistants  has  on  consumer  brands.  The  main contribution  is  to  reconcile  existing interdisciplinary  literature and  review how  voice  assistants may alter  market dynamics as emerged during in-depth interviews with 31  AI-aware executives.” ]

VA Devices leverage many different types of emerging technologies, top most important ones being ; Text o Speech recognition (TTS) , Automatic Speech Recognition ? (ASR) , and Natural Language Understanding (NLU) .

Va Device most popular functions with user are playing music, controlling smart home appliances, providing weather information and current time, Setting alarms, and general facts and questions asking to be answered.

[ “A report suggests that 21% of U.S. smart speaker owners have purchased entertainment such as music or movies, 8% household items, and 7% electronic devices (eMarketer, 2019). Meanwhile, Alexa’s users can order items like household products and fresh produce from a local Whole Foods and receive delivery within two hours.” ]

Characteristics of Voice Assistants (VA) :

Natural Conversation ; VA logs memorizes past Conversations / Questions leveraging topics of interest to better suit results for anything that may be asked of it.

Context Awareness ; VA is aware of its context in the home it works in. (Identifies the users Location, Time/Date, purchasing History, and preferred User preferences )

Self Learning ; Unsupervised Learning, Constantly Learning and Updating its Algorithms (Neural Networks)

VA devices are still only able to process single commands at a time , vs. many tabs and screens open at once when scrolling the web. (This will change is future)

I do not believe voice commerce will be become more popular or ever truly adopted until the option is available to have VA devices display a screen or project a hologram providing the visual aspect desperately required in the online shopping experience. People want to be able to visually see the product before the making a purchase. Due to short term memory problems and a Reduced attention Span problem in our population, Users reported having trouble remembering all the items in there basket they were trying to order at one giving time making the experience frustrating and a lot harder. 

The company / Agency behind the physical VA device being sold is the the sole Market Mediator between the customer and the product, these agencies are the ones controlling the ‘Black Box’ algorithms behind the search result recommendations, as a result of this the ultimately become the primary controller of the entire market available to users. (Whichever company creates the Algorithm code for the device holds all the control)

[ “The  strategic  goals  of  the retailer, merchant, advertiser, and voice assistant itself, may differ from those of end-user … For instance, a  VA  might recommend a private label (AMZN Basic Product) over a consumer brand following the retailer’s objective to swiftly grow its shares in a specific product category.” ]

Recommendation systems have proven to be biased in there results leaning heavily towards marketing there own Private label products over other competitors in the market. With this type of control of the market over the end User it is becoming drastically easier to monopolize any industry the developers want by eliminating the opportunity for any other competitive brands to ever be discovered.

How can we grow and discover new topics or brands is ‘Black Box’ A.I. algorithms make all our decisions for us ? 

These systems need to be transparent and regulated to protect the Market and the Consumer from being fed biased and misleading results to best suit the companies profits and personal interests.

[ “Throughout the  collection  of  significant  volumes  of  personal  and  behavioral  information,  VAs  can  push  users  to  automate repurchase,  for  instance,  via  “subscribe &  save”  promotional  activities… In the context  of purchase automation, consumers might have  aspirational  preferences  that  differ from  the  preferences  suggested  by  their past  behavior.” ]

An example of how this could happen and hinder the end consumers experience ;

[ “… in the case of an environmentally aware person who wants to use less bottled water but is regularly reminded to buy plastic bottles. The inherent tension between the actual-self and the ideal-self represents a boundary for those consumers who follow VAs’ suggestions to automate repurchases.” ]

By relying on the suggestions of your VA Devices more and more moving forward, one can make the argument that the VA and consumer experience today is shifting from simply influence; into a steadfast dependency to selection of results they are offered.

Search algorithms represent the gatekeepers for modern companies and retailers.