How to Make Better Talent Decisions with AI-Driven Predictions [Webinar Recap]

Artificial intelligence is a buzzword in almost every field, and recruiting is no exception. In the last year or so, it seems like nearly every recruiting technology vendor has added “AI” to its platform.

But with buzz comes mixed messages. Some days you might hear about how AI is a bias creator; on others, you’ll hear about how AI can remove bias.

Recently we hosted a webinar to cut through some of this confusion. Led by our SVP of Game-Based Assessments Clemens Aichholzer and Chief IO Psychologist, Dr. Nathan Mondragon, we explored where (and how) AI should be used in recruiting.

Missed the webinar?
Watch it here

The Potential for AI in Recruiting

Clemens started off with a brief overview of AI, and gave us a simple definition: AI is making better predictions (usually based on large amounts of data).

There’s huge potential for AI to assist with things like talent identification, task automation, and candidate feedback. Properly leveraged, AI can assist in the human decision-making process: increasing new hire quality, decreasing time to fill, and removing bias.

That said, recently a prominent tech company was scrutinized for training an AI to screen resumes - and receiving extremely male-biased recommendations as a result. In this case, AI seems to be a bias creator.

But just last year we learned how Unilever used AI to hire it’s most gender and ethnically diverse class of graduate recruits to date. What’s different? The key, Clemens explained, is combining AI with IO. In other words:

The key is combining leading edge data science (AI) with proven people science (Industrial Organizational Psychology).

You can’t just throw artificial intelligence at a problem and expect great results, particularly in a highly nuanced domain like recruiting. Data scientists today are not educated or trained to take bias into account, or to evaluate their models for adverse impact. If you want to apply AI to recruiting, it’s critical to work with a vendor who understands the space and has experience eliminating any potential pitfalls.

That’s why it is essential to couple great data science (AI) with proven people science (IO Psychology) when applying artificial intelligence to recruiting.

Proven People Science: Pre-Hire Assessments

Pre-hire assessments - created by trained IO psychologists - have been used for decades to assist in making scientific, validated talent decisions. Pre-hire assessments are designed to predict employability: how a candidate works with people, how they work with information, their personality & work style, and their technical skills. Legacy pre-hire assessments are question-based. They take the form of a 60-120 minute survey, completed by candidates as part of the hiring process. While scientific and validated, this legacy approach to assessment has several limitations.

• Lengthens time to hire. Adding additional steps (several tests) to the hiring process lengthens time to hire.
• High candidate dropout rates. Many candidates will prioritize pursuing other job opportunities if they are confronted with a lengthy test.
• Potential for bias. Legacy assessment test cannot be altered to remove adverse impact without significantly impacting their ability to predict performance (which is the point of the assessment).
• Cannot keep up with rapidly changing job types. Refreshing a legacy assessment requires completely overhauling the test, effectively starting from scratch. While this might be acceptable for jobs that stay the same year after year, it does not reflect the new reality of the rapidly evolving workplace.

As Clemens explained, while the internet allowed these legacy multiple-choice tests to be ported online, it was only until recently that these limitations could be overcome. This is where AI comes into play.

Artificial Intelligence + Proven People Science

Today, you can assess candidates with the same scientific rigor and validity in a more expedited, candidate-friendly, bias-mitigated way through:

1. Video-Based Assessments. Candidates’ performance in an on-demand video interview is evaluated with AI.
2. Game-Based Assessments. Candidates’ performance in game-based challenges is evaluated with AI.

These new assessment methods come with all the benefits of legacy assessments (scientific, validated), and overcome their limitations.

• Decreases time to hire. Combining video-based assessments with game-based assessments results in a single, 20-30 minute assessment experience. Recruiters can then watch candidates’ interviews and forward the best to hiring managers - replacing the resume screen, the assessment, and the phone screen.
• Keeps candidates engaged (low dropout rates). AI-driven assessments usually see completion rates over 90%, Clemens explained.
• Remove bias. Due to the large amount of data collected in recorded video and gameplay, data points that contribute to adverse impact can be removed without negatively impacting the predictive power of the assessment.
• Quick and constant updates. Again, due to the large amount of data collected in recorded video and gameplay, AI-driven assessments can be continually updated and refreshed to reflect changing job needs.

See how this combination of artificial intelligence and IO Psychology works in practice. Watch the full webinar:

Watch the webinar

In the webinar, you’ll also get a deep dive into HireVue’s bias-removal process, as well as the answers to the following questions:

1. We have an assessment with pretty low completion rates. What percentage of candidates do you generally see complete the assessment?
2. Do candidates receive any feedback? What do the reporting capabilities look like?
3. Can the games only be completed on a mobile device? We have a lot of candidates who only have desktop.
4. What if the needs or requirements of the folks I hire have changed, is there a way to assess for talent changes the assessment doesn’t measure yet?