Is AI the answer to inclusion recruitment?

Human biases in recruitment are well documented. Many organisations are turning to Artificial Intelligence (AI) to eliminate conscious and unconscious biases influencing the hiring process, but is such ‘smart’ technology really the solution to a very human problem?

I can see how AI would appeal to many organisations as it can help manage a higher volume of applicants with the added benefit of reducing human bias, resulting in a more inclusive and cost-effective recruitment process.

Sounds perfect, right?

If only it were as straightforward as that. Organisations attempting to ensure inclusivity when hiring by adopting AI solutions have often found the software to be flawed. They introduced similar biases to those they were trying to overcome. AI operates from the data it is fed. If that data is biased in the first place, intentionally or otherwise, any inadvertent prejudice will be compounded and perpetuated, potentially resulting instead in a lack of in-house diversity.

Mckinsey tells us that;

Bias can creep into algorithms in several ways. AI systems learn to make decisions based on training data, which can include biased human decisions or reflect historical or social inequities, even if sensitive variables such as gender, race, or sexual orientation are removed.

A good example of AI bias can be found in Amazon’s history.

Amazon scrapped its AI recruiting tool because it uncovered a big problem: their AI recruiting tool did not like women.

Amazon’s system taught itself that male candidates were preferable. It penalised resumes that included the word “women’s.”

Amazon edited the programs to make them neutral to these particular terms. But that was no guarantee that the machines would not devise other ways of sorting candidates that could prove discriminatory.

Ultimately, the program was disbanded as being unreliable.

(Source: Reuters)


So what’s the answer?

Truly inclusive recruitment cannot be fully automated or replaced with AI technology. It is more effectively enabled alongside human interaction to facilitate well-thought-out processes and approaches.

Back to news