Raising Concerns of AI Video Interviewing: A Study in Experience Architecture

Overview

While interviewing for a post-graduate job, I encountered a video interview process that was powered by Artificial Intelligence (AI). Many concerns, such as accessibility and usability, arose for me both during and after the interview, and I began to investigate the technology for user perceptions and experience as well as concerns of accessibility, ethics, and usability.

Role

For this project, I was the sole researcher and writer. I received feedback from Dr. Liza Potts and my peers as part of her Foundations of Experience Architecture course as well as the editing team at UXPA Magazine.

Methods and Tools

Participatory Research and Observation, Reading/Viewing

The Goal

My goal for this article is to further investigate the new and upcoming technology of AI-powered video interviewing, where job candidates are assessed via video submissions and an algorithm. Although not much research exists about this technology, I at least wanted to raise some concerns I personally had as well as some user perceptions to inform employers of some cautions associated with implementing this technology.

Approach

After my experience of interviewing with AI video interviewing technology, I became concerned particularly with the platform’s accessibility because of my research interests stemming around technology and accessibility. As I began to make note of these concerns, I reached out to Dr. Liza Potts and my classmates in WRA 891: Foundations in Experience Architecture to obtain their initial thoughts. Through those conversations, concerns over user experience and ethics were raised as our class often discusses the ethical impacts of AI and algorithms.

Therefore, I began to investigate this technology for accessibility, user experience, and ethics. My initial plan for this project was to frame it in a “pros and cons” perspective as the neutrality would prevent the piece from becoming a hit on the technology itself. Ultimately, I strayed from this idea because there was limited scholarship on either side so I opted to instead present any findings I did have so that companies and job candidates could make a (more) educated decision as to whether or not they choose to implement or use this technology. In that aspect, I was also able to avoid bias as the only scholarship I could find was produced by researchers affiliated with the technology.

Through this initial inquiry, as a student in rhetoric and writing, I began to research the platform I had experienced—HireVue—and others’ experiences with it. The overall experience was new to my colleagues and therefore I took to online discussion platforms like Reddit for candid discussions of users’ experiences. After scouring a few Reddit threads, it became clear that there was a divide between the opinions of the job candidates and the companies that implement this technology. Essentially, I found that the majority of users (based on these threads) felt the technology was impersonal and questioned whether or not this algorithm could accurately predict if they would be a good candidate for the job in question. On the other hand, people in management positions liked the streamlined approach that HireVue offered which allowed them to assess a wider variety of candidates with less time and fewer resources.

In writing the article, I struggled with appropriately identifying my audience and had to tailor this piece to suit two different audiences. First, it started with a more academic and scholarly tone for my course final. Then, to be published in UXPA magazine, I had to unravel this tone so that it would match the magazine. To do so, I had to be extremely diligent in ensuring the content was readable and clear. At times, this meant defining jargon or concepts. Additionally, I worked with the magazine’s editorial team to adhere to their format and citation guidelines.

Ultimately, my article was published in the March 2022 issue of UXPA magazine and effectively presented my findings in a way that was clear and concise. I was not able to fully investigate the concerns I had, but I was able to raise them and open the door for more conversations about AI interviewing technologies and their ethics, accessibility, and user experience. There is still much research that needs to be done; especially if this technology is going to become increasingly prevalent.

Deliverable

Kayden_AI_Interviewing

Outcome

Ultimately, this article was accepted for publication with UXPA Magazine. Although the article itself only raises claims and doesn’t deeply investigate them, it paves the way for further research before use of this technology does (or doesn’t) become widespread.