The article from the Washington Post talks about HireVue, the recruiting-technology firm. HireVue’s AI-driven recruiting system is widely used for the recruiting process of well-known companies such as Unilever, Hilton, vodafone, Oracle, etc. (https://www.washingtonpost.com/technology/2019/10/22/ai-hiring-face-scanning-algorithm-increasingly-decides-whether-you-deserve-job/)
In case you are not familiar with HireVue (like me) and since the article is substantially long, let me summarize things about HireVue.
HireVue launched its AI assessment service in 2014 as an add-on to its video-interview software, which more than 700 companies have used for nearly 12 million interviews worldwide. The company’s revenue and a full list of clients are not disclosed.
2. Product; Video interview analytics
HireVue records job candidates’ interviews and analyzes their facial movements, word choice, and speaking voice. It uses a proprietary algorithm to compare candidates’ assessment results to the results of high achievers of the companies. HireVue claims that it can measure how candidates score on various competencies, which may include communication skills, compassion, empathy, and the ability to work in a team or make prudent decisions.
3. The benefit of the product; HireVue claims;
HireVue provides time and cost saving to employers. For example, Unilever said it saved 100,000 hours of human recruitment time in a year by using HireVue. It also claims the system generates more consistent and predictable selection results without human biases.
HireVue also provides candidates better experience and accessibility since it leverages digital devices.
4. The algorithms and how it analyzes and selects;
HireVue doesn’t disclose much about algorithms both to protect trade secrets and because the company doesn’t always know how the system decides on who gets labeled a “future top performer.”
The article tries to make a balance between proponents, namely executives of HireVue, and some opponents, although it leans toward skeptical view overall.
I do agree with the skepticism. Although consistent, structured interviews and analyses can help companies mitigate the issue of human error and biases, it is not clear how accountable the system and algorithms are.
First and foremost, the unique proposition of the system, analyzing facial expression, depends on a false assumption of “universal facial expression”. While facial expression can tell many things about a person, not everyone and every culture share the same facial expression. For instance, people from East Asia (like me) don’t have much facial expression to read when they have a conversation, especially in a formal situation like an interview. The facial expression doesn’t tell us context as well. As an example from the article, a scowling can appear when people are not only angry but concentrating really hard or confused.
Second, it is not clear how HireVue’s assessment can be connected to that of high achievers. Is there something unique as well as common about facial expression, word choice, and speaking voice of high achievers?
Third, HireVue might select the same type of candidates who share similar culture, background, etc., resulting in less diversity and discouraging non-native speakers. To make things worse, as the article mentioned, the recruiters like the platform because it saves tons of time for them. The recruiters don’t have interest and time to look who wasn’t selected by AI, once they decided to use the system.
Sarah Smart, Hilton International’s vice president of global recruitment, said the system has radically redrawn Hilton’s hiring rituals, allowing the company to churn through applicants at lightning speed. Hiring managers inundated with applicants can now just look at who the system ranked highly and filter out the rest: “It’s rare for a recruiter to need to go out of that range,” she said.
I am sure the AI and data analytics will be the main tool for future recruiting and hiring. However, I want that to happen when we have more transparent and scientifically proven solutions.