Last month at the Unleash show in Amsterdam, I spoke to Clemens Aichholzer, MindX's founding CEO and now SVP of Game-based Assessments at HireVue, about the combination. HireVue likes to call the resulting blend of the two different AI-powered assessment techniques a "hiring intelligence" solution, he tells me:
We can use games to measure traits that are most suited to be measured with games. We can use video intelligence for those traits that can best be measured with video.
The game-based approach works best for measuring cognitive traits. This is mainly because they're really just applying a game dynamic to what used to be tested in a traditional mental ability test. The game setting allows the test to quickly find someone's level and test their potential in a specific characteristic.
Cognitive traits are relevant to qualities such as logical thinking, attention, visuospatial acuity and related fields such as computational thinking and planning, says Aichholzer. These can be supplemented by specific skills tests, for example coding challenges for use in software developer recruitment.
Video assesses social traits
Analysis of video on the other hand allows to measure characteristics such as personality and social traits, and thus teamwork capabilities, empathy and so on. While in the early days HireVue simply enabled a remote interview — and many clients still use it in this mode — the company has since added AI tools that allow it to make an automated assessment of the candidate's behavior during the interview.
For high-volume requirements, candidates these days may find themselves responding to a series of recorded video questions rather than a live interviewer. HireVue doesn't yet offer the option of interview by software avatar, but it may come in the future.
Video techniques come to the fore when recruiting for roles that require a lot of personal interaction, such as a salesperson, whereas when hiring for more of a back-office role such as a data scientist or an auditor, the emphasis is on games-based assessment. But for most roles, a combination of the two works best. Even software developers need social skills, Aichholzer points out:
At the end of the day, even if you are a coder or a developer, you're not working in isolation. You still need to meet deadlines. You still need to communicate your work streams to your team members. So we found that just giving them a coding challenge in itself is by far not as predictive of their job performance — compared to an integrated assessment that really looks at what we have identified as important components for the profile of, for example, a software engineer.
To help customers apply the right combination, HireVue recently launched its first pre-built assessment offerings, tailored to specific types of job. The first ones available are for call center agents, sales representatives, retail associates and software developers. These packaged assessments can be put into use within a couple of weeks, says Aichholzer, rather than the twenty weeks or more it can take to build a custom set of tests.
Can we trust the machines?
The objective of this automation is not to replace the HR professional, but to help them be more effective, Aichholzer emphasizes.
We see the software that we are providing really to empower, unleash or augment the position of the HR professional. We help them in a high-volume scenario to surface the most promising candidates. And then the HR professional can focus on this subset of candidates that have already been assessed through an objective, data-driven, rigorous, digital approach.
It may sound depersonalizing for candidates to be assessed by online tests and recorded interview questions, but Aichholzer points out that it's a much better way of evaluating suitability for a role than filtering people based on what's on their CV. Instead of winnowing out candidates based on their paper qualifications, the automation allows recruiters to look at a wider and potentially more diverse candidate pool, he argues:
It's tempting to say, if someone has already been prescreened — if they have gone to a good school or they've done a good degree — they must be smart. Basically you outsource the assessment of your candidates.
What we can offer is almost a more democratic process ... The problem that employers sometimes have is that they artificially narrow their applicant pool. They could cast a much wider net if they had a reliable assessment and — in a way have a fairer process — basically identify talent in spaces that are not as conventionally obvious.
Can we trust the machines not to have in-built bias of their own? Aichholzer says that ensuring the HireVue model doesn't build in bias is something the company is paying keen attention to:
I personally think we are almost in the best position to do so. It takes a multi-disciplinary team that doesn't only consist of machine learning experts, but then also consists of HR professionals, industrial and organizational psychology experts and potentially even an ethicist. Put it all together and have a cross-functional team to develop this AI-powered assessment hiring solution. That's something that we feel very comfortable doing because that's our bread-and-butter.