cu Reed Snider

A Look At Artificial Intelligence

Researching how far personalization can go before it is too personal and how people look for and receive help in today's digital world.

In this project I worked with a leading artificial intelligence company that creates, among other things, chatbots for pizza companies. The project was to help the company discover more about how this growing AI technology can fit into our lives. I conducted this study using a card sorting activity and in-depth user interviews.

One of the primary goals in this study was to find out how far personalization can go before it is too personal. The question is what amount of information is excessive or intrusive for a computer database to know about a person and why do those people hold such beliefs? The secondary goal was to explore how people look for and receive help online and in the store and what methods they use to solve their problems.

location identification notification

From this research emerged three themes:

  1. Help and privacy are not correlated. This means that if something is too intrusive to one’s privacy it will be avoided even if it is considered extremely helpful.

  2. Unsolicited help is a sensitive matter. You must find out who the audience is and properly evaluate when and how the help is being given. When people receive help that they did not ask for it can be very alarming. Not knowing where that information was gathered from is another layer of worry that can cause people to feel uncomfortable. People want to be in control of their personal data and when they cannot determine how their information is being used many people begin to panic. These type of people are often suspecting the worst-case scenario like a security breach or unauthorized data mining.

  3. Informed consent can build trust. When people realize where the data is coming from, such as an ad that uses previous search history, they worry less. When people begin to understand the way in which their data is being used and agree to those terms, that transparency creates a sense of acceptance and people will be more comfortable using your product. People do not trust companies that skirt away from the truth and this applies directly to data mining and the use of personal data.

Card Sorting Activity

In this exercise please order these scenario cards from what you consider to be the most intrusive and least intrusive to your privacy. Next, I will ask you to order these in terms of being the most helpful to least helpful.

Questions Shown On Cards:

  • You receive a notification about a local restaurant that just opened near you.
  • You receive an email from your bank to recommend their services; it comes with a picture of you and a friend in it.
  • While shopping at the grocery store your phone sends you a list of relevant coupons.
  • You call an airline to change your flight and they answer the phone saying, “Hello, [your name]”
  • You walk into a bar and a bartender you’ve never met already knows your favorite drink and where you like to sit.
  • Facebook sends you a pre-made post with you and a friend encouraging you to post it to your profile
  • You receive an ad for hiking socks after searching for local hiking trails

Probing Questions:

  • Can you tell me about how you ordered these cards?
  • Let’s move down the list, can you tell me about the next scenario and why you placed it there?
  • I saw you place “x” at the end/start of the list, why is that?
  • When do you consider something to be intrusive?
  • Do you see a point between any of these cards that you consider a point of acceptable versus unacceptable?
  • Why?
    • Highlights
      • Deep diving into qualitative research
      • Facilitating one-on-one moderated studies
      • Digesting and consolidating research into a consumable report
Reed Snider contact

This website was coded by Reed Snider