SHL’s Video Interviewing Solution Unites the World’s Workforce

Video interviewing is indisputably a powerful and valuable new hiring tool, freeing both candidates and talent acquisition teams from the constraints of schedules and travel, and potentially opening the interview process to many more candidates. However, to effectively and efficiently manage more interviews, organizations also need to scale video interview scoring.

Artificial intelligence (AI) shows promise as being a powerful decision aid and brings even more standardization to the process. AI is fair and useful if it has a foundation of science underpinning the technology.

Before SHL’s video interviewing solution there were two kinds of AI video interviewing tools:

One kind of tool analyzed the videos using some off-the-web AI-based facial recognition analysis. The tool would spit out a bunch of scores, but with no basis or evidence to say that the results are job-related, or that they are predictive of hiring decisions or job performance.

The other kind is a tool that trains its AI algorithms to mimic the current hiring process or predict top performers on the job. The problem here is that the approach has no theoretical basis in assessment science. The algorithm could simply propagate the biases in the current system. For example, if a company is under-selecting or rating women down, so would the algorithm.

Clearly, neither of the above approaches work. So, what is the solution? Not use video interview? Absolutely not… Science comes to rescue!

SHL uses validated AI algorithms to score candidates on job-related competencies. Science informs which competencies we assess with our AI technology based on the job requirements.

Competencies assessed can include areas such as social skills, confidence, and work management. We can also assess domain skills and job knowledge (such as Java skills) based on the content of responses provided by candidates.

We use content valid questions and develop Natural Language Processing (NLP) based models to measure responses using the text of their answers. Customers can configure different scoring in the video interviews they use for different roles, and then combine the relevant competency scores in an automated scoring model. This combination happens in the same way as with traditional assessment.

Science informs which competencies we assess with our AI technology based on the job requirements.

In order to ensure that our AI algorithms—and in fact all of our assessments—are fair and do not exhibit bias, SHL examines between-group score differences both during product development and after implementation of our algorithms with clients. In accordance with professional standards such as the Uniform Guidelines on Employee Selection Procedures, we examine scores for evidence of adverse impact against protected groups. If an AI algorithm appears to exhibit adverse impact, we look at the constituent features in the algorithm to identify and remove those features that are producing the unintended score differences.

Using science to guide AI makes video interviewing tools immensely powerful. By assessing job-related competencies, using empirically validated AI scoring, and monitoring algorithms for fairness/bias, we can harness the power of AI for great talent decisions and make it a win-win for all.

Learn more about all of the hiring tools that we have to offer.