top of page

AI Discovery Research

Amidst the growing prominence of Artificial Intelligence, I contemplated the present and future impact of AI on user experience research. I approached my supervisor at Verizon to inquire whether our team has taken into account the implications of AI in our domain. This inquiry initiated discovery research that aimed to comprehend the nature of AI and its potential implementation in Verizon's research practices. The study comprised two parts: the first was to explore the existing types of AI and their applications across various industries, as well as examine AI's use cases in user research and ethical concerns; the second was to identify potential use cases for AI within the Verizon research team and understand the perceptions regarding how AI can influence research quality and practices.

Tools



    Slack - Slack was utilized to recruit message members of the market research and user experience research teams for recruitment for the 2nd part's interviews.


    ​Google Meet -  I utilized Google Meet for remote, moderated usability interviews with members of Verizon's market research and user experience research teams. Demographics of users consisted of a mix of gender, race, positions within research teams.


    Google Office Suite - I utilized Google Office Suite to write test plans and to create findings presentations.


    Miro - I utilized Miro for notetaking during sessions and keeping track of insights and themes.


    Databases (EBSCO, ProQuest, College of Southern Nevada Library, Google Scholar) - I utilized these databases to find scholarly articles and articles written by UX and AI professionals within the first part of the study to understand existing types of AI, their use cases, and ethical concerns of AI.

Timeline

Like all studies submitted to the Rapid research team, each park of this study needed to be completed within a week's time. Moderated interview sessions were 30 minutes each.   Below is the weekly timeline detailing the each part of the study.

Part 1

Part 2

Day 1

  • Part 1 begins.

  • Send out invites for Kick-off meetings with stakeholders.

Day 2

  • Meet with stakeholders during the Kickoff meeting to walk through objectives

  • Write the test plan after kickoff.

  • Submit test plan to manager by end of the day.

Day 3

  • Begin searching through research databases and web.

Day 4

  • Continue searching through research databases and web.

Day 5

  • Continue searching through research databases and web.

Day 6

  • Synthesize findings and then create finding presentation

Day 7

  • Hold a Read-out meeting with stakeholders to present findings.

Day 1

  • Part 2 begins.

  • Send out invites for Kick-off meetings with stakeholders.

Day 2

  • Meet with stakeholders during the Kickoff meeting to walk through materials, and objectives

  • Write the test plan after kickoff.

  • Submit test plan to manager by end of the day.

Day 3

  • After manager reviews test plan, researchers send the test plan to stakeholders for further approval

Day 4

  • Finalize changes made to test plan

  • Testing begins at 1pm EST

Day 5

  • Testing with participant continues all day

Day 6

  • Synthesize findings and then create finding presentation

Day 7

  • Hold a Read-out meeting with stakeholders to present findings.

Key Takeaways

Part 1



    There are several distinct types of machine learning and vary by rule sets and human interventions.



    There are 2 main types of AI: Hand-Crafted and Machine Learning AI.


    Supervised learning, unsupervised learning, Semi-Supervised learning, and reinforcement learning are machine learning types and are different ways of categorizing data within an algorithm.




    Privacy, surveillance, behavior manipulation, bias, and transparency are ethical concerns of AI



    Lack of regulation amongst AI systems leaves ethics to developers who create the AI and to users who utilize/input data into AI.




    AI currently helps and could further help UX researchers plan, conduct, and synthesize.



    UX Researchers are already using AI through UserTesting and UserZoom platforms to source participants, transcribe sessions, and analyze test results.


    Machine learning systems such as ChatGPT can help researchers with script writing, persona creation, generating feedback users might say, and generating comments and suggestions.


    When inputting research data into AI systems that are not owned and maintained by Verizon, researchers should avoid inputting sensitive information of users and Verizon.



Part 2



    Users have used AI before in some capacity whether they knew it was AI or not.



    Only one user claimed that to the best of their knowledge that they had never utilized an AI tool in the past.


    The remainder of users were aware that AI was being used across the web and that they have used AI at some point whether it was an active choice or not.




    Users feel that AI tools currently don’t have capabilities to support research significantly.



    At AI’s current state, users feel that AI cannot analyze data in the same way a human could, because it cannot replicate human thinking.


    Users feel that AI will advance enough in the future and will benefit research practices, but right now it simply does not have strong enough capabilities to make an impact.




    Users feel that AI could be dangerous if not regulated and utilized properly.



    Some have general concerns that AI will eliminate many jobs; however, they don’t seem concerned about the elimination of research
    jobs, instead feeling that AI will change the nature of research.


    Bias and transparency were the biggest concerns amongst users, because AI tools that could help researchers in the future may be full of bias, emphasizing the importance of transparency to regulate the amount of bias within AI tools.



Challenges



    Lack of secondary sources: Due to the novelty of AI, especially machine learning, there is a scarcity of peer-reviewed scholarly articles on the subject. However, I did manage to find several valuable peer-reviewed sources, albeit I had to heavily rely on web articles authored by UX and AI professionals. In order to ensure credibility, I meticulously researched the reliability of these articles, resulting in the exclusion of numerous sources that did not meet the desired standards. Ensuring the quality of my sources took up the largest chunk of time during part 1 of this project.


    Recruitment: This study deviated significantly from our typical rapid research approach, which involves recruiting customers or potential customers. In part 2 of this study, I faced recruitment challenges as I needed to interview fellow researchers within Verizon. Surprisingly, there was limited interest among researchers to participate as research subjects, resulting in a small pool of potential participants. Unfortunately, half of the initially interested group eventually withdrew, leaving me with a total of 7 participants. Despite this setback, I was able to gather valuable insights on the perspectives of these 7 participants regarding artificial intelligence.

Location

Austin, Texas

Email 

Social

  • LinkedIn
bottom of page