Many organizations today are using AI algorithms to identify and qualify potential job candidates. A 2022 study, “Hidden Workers: Untapped Talent” by Fuller et al, Harvard Business School discovered that these algorithms not only overlooked quality candidates but also potentially discriminated against specific groups such as immigrants and veterans.
“Experts warn that as AI becomes smarter, it could also become biased towards certain candidates.” Laura McQuillan, CBC News, Jan 19, 2023.
THE PROBLEM: OVERLOOKING QUALIFIED CANDIDATES AND DISCRIMINATING AGAINST PROTECTED GROUPS
Overlooking qualified candidates and discriminating against protected groups has recently resulted in class action lawsuits. For example, in 2018 Amazon was forced to drop their AI algorithm because they discovered the algorithm began discriminating against females. The major issue we are seeing with AI is that organizations are not always aware of the factors that are loading on the algorithm to identify qualified candidates. Furthermore, with machine learning constantly changing the factors and how they are weighted within the algorithm, it could be OK when checked at one point in time but then change to become discriminating at another.
Qualified candidates are also overlooked if there is a gap in the employment history because the machine has learned to disqualify any candidate who has a period of time where they were unemployed, regardless of the circumstance behind it. At SMG, we recently interviewed an experienced training consultant who had a 2 year gap in her employment history because her mother became ill and she decided to take a hiatus from work and care for her – the AI algorithm wouldn’t be able to pick up on the reason for the gap but simply discard her because of it.
In todays remote and hybrid work environments it is very easy to increase the flow of candidates to job postings if the position is advertised as “remote”. However, the quality of candidates starts to become questionable due to the fact that many applicants are simply applying to be remote, and not because they are qualified or even interested in the position. One of our clients recently changed their competitive sales compensation model from 100% commission to a salary plus over ride. The change in compensation significantly increased the flow candidates, however the applicants were mainly unqualified and only interested in the salary component. The majority of candidates didn’t know how to prospect and were considered more “account managers” rather than “competitive sales professionals”. These are 2 examples of how AI would begin to shift the hiring model and potentially discriminate or overlook quality candidates due to all the “unqualified noise” it has coming through it’s system.
In both cases there was a tremendous administrative load on the recruiters filtering through the flow to find any qualified candidates. As a result, rather than increasing the efficiency and effectiveness of the recruiters it had the opposite effect.
THE SOLUTION IS EASY: LOAD VALIDATED SCIENCE ON THE ALGORITHM
To maximize the use of AI in the talent intelligence area simply load a proven, predictive, validated algorithm on the AI platform. The algorithm must be both predictive of performance & retention and proven to be non-discriminatory.
In summary, AI is revolutionizing talent acquisition and talent management processes. In order to help increase the efficiency and effectiveness of AI and avoid a class-action lawsuit, it is important to add science to turn your Talent Intelligence into Predictive Intelligence.
Based on almost a half century of research and validation, SMG has proven algorithms that identify DIVERSE, ENGAGED & PRODUCTIVE CANDIDATES WHO WILL BE RETAINED BY AN ORGANIZTION. In addition SMG performs benchmark studies to help our clients build customized, predictive algorithms that can be integrated and loaded on any AI platform.
For more information on how we can help turn your Talent Intelligence into Predictive Intelligence contact us at email@example.com or visit www.selfmgmt.com.