AI

Artificial Intelligence (AI) Is All The Rage – But Will It Help HR?



The quick answer is….apparently not!

At Peoplogica we love technology, we love people analytics and we love innovation, but from day one we have been a little wary about the use of AI in the candidate selection process; in particular using AI to screen applicants.

The reason for this wariness is that we have already witnessed the failings of automated resume scanning, whereby professional resume writers optimise candidate resumes using keywords, irrelevant of whether the keywords used are related to that particular candidate’s experience and qualifications. And, of more concern, where potentially great candidates are not even considered by a human because they don’t know how to “game” the system.

So what positives could AI potentially bring to candidate selection? Well, number one is that it will make it so easy for HR to develop a shortlist, number two………….umm………….well it will make it easier for HR.

Okay, let’s review the potential negatives. Actually, we don’t have to guess what the potential negatives are because Amazon has been trialling AI in their candidate screening processes for a few years, and this is what they found (full article):

  1. AI discriminates against women
     
  2. No matter how hard the engineers tried to stop this discrimination, the AI engine found a way to discriminate against women
     
  3. AI downgraded your suitability if you attended an all-women’s college
     
  4. After 500 iterations that were each trained to understand 50,000 unique terms, the AI system still found a way to discriminate against applicants
     
  5. As the project progressed, the AI system regularly recommended unqualified candidates for no apparent reason

Now, maybe some of the above was due to the tech sector being traditionally male and therefore AI is using the past as a basis for its artificial thinking process (a little bit like human unconscious bias), but doesn’t that pose an even greater problem regarding diversity in the future. There are many roles that are/have been gender dominant (such as aged care, construction, child care, firefighting, HR, IT, auto mechanics, mining, nursing, receptionist, teaching etc) but as humans we are looking forward to what balance we believe we need in the future. This is because we have free thought, we do not always allow ourselves to be dictated by the past. Apparently this is not the case yet with AI.

But in the end my greatest reservation about using AI in candidate screening is that there is far more to attracting, selecting and retaining great employees than the two dimensional world of resumes. Selection should be thought of as a four legged chair. To consistently select future high performing employees you need to focus on:

+ The Resume, and

+ The Interview, and

+ The Psychometric JobFit Assessment, and

+ The References

None of the above should account for more than 40% of the decision, and this is the problem with AI in resume screening. It represents 100% of the decision making as to whether a candidate’s application is successful in the sorting process and therefore whether it is actually looked at by a human.

Our advice: Absolutely investigate and pilot AI technology, but do it because it will deliver a superior result not just because it makes the screening process easier. What may seem like an avenue to save costs and reduce time could in fact end up causing the exact opposite if you consider the substantial imposts caused by bad hires and high early failure rates!!

Written by Mark Purbrick

Leave a Reply

Your email address will not be published. Required fields are marked *