The problems with using AI to establish a diverse workforce

One key challenge that Elliott identified in regards to recruitment was finding ideal talent while keeping diversity in mind.

“Recruitment is tough,” he explained. “You’re often in a rush to fill key positions, and unfortunately, the people with the right skills are always highly sought after.

“Many recruiters get hundreds of applications for their jobs, which means that hiring processes can entail sifting through endless applications to find the right person for the job.

“On top of all this, hiring processes are always trying to provide a good candidate experience that pays a keen eye to diversity.

“Today, candidates expect a better experience when applying for a role. They expect efficiency in the application process, efficiency in different stages of assessment, and efficiency with finding out whether they’ve been successful or not. But, crucially, candidates also expect to be fairly evaluated, regardless of background, colour, sexuality or nationality.”

“The recruitment conundrum: Assembling effective and diverse teams”

Recruitment is one of the biggest challenges facing employers, who are looking to hire the most diverse and effective teams. Read more

While the entrepreneur stated that AI can be helpful when implemented into the recruitment process within companies, he says that when it comes to the matter of diversity, human intervention can’t really be replicated.

“On paper, implementing AI into hiring processes is proving an appealing time-saving tool for businesses,” he said. “What many are overlooking, however, is that it also leaves them extremely vulnerable to a couple of glaring weaknesses, namely fairness, diversity and personality, which are difficult to achieve through data.

“It takes a conscious human output.”

Potential bias

Elliott went on to explain that while AI can possibly cut down time when it comes to searching through applicants for jobs, such systems may not be effective when bearing diversity in mind.

“In recent years, tech providers have started to release new products and features that claim to use AI as a tool to help recruiters,” he said. “They do this by taking the front load of initial candidate shortlisting, before then using AI software to order the long list of applicants and show busy recruiters which candidates are worth them taking a personal look at.

“With VR, Openreach asks job applicants to climb telephone pole ‘virtually’”

In the hunt for 1,500 new trainees, Openreach will use virtual reality headsets to show job applications what it’s like to be a field engineer. Read more

“Sounds good, right? On the face of it, yes. When you’re busy and there isn’t enough time, a machine doing the legwork sounds ideal.

“However, AI systems work by analysing pools of past data, finding patterns and drawing conclusions, before then musing those patterns to make predictions about future data. This is all fine if these initial patterns of data are based upon a history of diverse candidate hires.

“Unfortunately, though, this simply isn’t the case. With past data sets heavily biased, especially around gender and race, AI is only serving to perpetuate this further by building a biased framework around the initial data it is given.

“The first generations of AI recruitment tools are likely to be the worst, and long-term, could have a huge impact on the diversity of teams.”

Case in point: Amazon

Amazon’s previous AI-operated recruitment system, which was scrapped earlier this month, could be seen as a notable example of a system that was shown to hold a gender bias.

“The learning point here is that employing AI in the selection and management of people is going to be tough, and fraught with danger,” Elliott explained in reference to the finding. “We’re going to have to work really hard from the outset to remove – and protect against – unwanted bias, which, to be honest, isn’t going to be easy.

“I think it’s a real warning sign for anyone excited about quick and near-term applications of AI where people are concerned.

“I, for one, want to be a laggard in this area. There’s no way I want any company I’m involved in to be in the first few waves of AI adoption in HR, as for me, the risks are just too high.”

A “broader approach” needed

The tech entrepreneur and author went on to explain that aside from ethics, recruiters must observe a much broader criteria in order to speed up the overall progress of workplace diversity.

“Digital disruption means ethics is playing catch-up”

There’s a shared responsibility between tech giants to try and create and utilise a critical ethical framework that assigns responsibility and accountability. Read more

“Recruiting more diverse teams is making slow progress because we’re still looking at the same, tired criteria,” said Elliott. “Whilst many recruiters have rightly adopted a fair appraisal process to applications, the criteria needed to even get to that stage is largely based on the same focuses of work experience and education.

“To really democratise how application processes play out, a broader approach needs to be applied, so that talent is judged against actionable skills, too; not just on a black and white evaluation of whether someone has a degree or not.

“Google proved several years ago that degrees and GPAs have zero correlation against future job performance, with the exception of a small set of people entering into their very first job after university or college.

“Yet, most companies still insist on a degree, rather than setting a standard for skills and experience which they can assess against.”

The Vital Role of Company Executives

In regards to the role of CTO’s and CIO’s in promoting diversity within their company, Elliott regarded the recent figure of 12% found by Wired and Element AI in reference to AI-orientated workers that were women as “not good enough, and it’s filtering through into the products we’re seeing today”.

“The future of being a woman in technology”

The glaring gender imbalance in the tech industry is there for all to see. Laura Malcolm, from Avanade, looks at the future of women in tech. Read more

“For example, we only need to look around at popular products in Siri, Alexa and Microsoft’s Cortana, to see the effects that a lack of diversity can have on reinforcing old stereotypes – with all three defaulting to female voices. Worse still, Amazon’s market leading Alexa is permanently a female voice.

“The key to assembling any great team is to have diversity at its core; providing teams with a range of unique perspectives to problem solving and innovation.

“Ultimately, CIO’s and CTO’s that place this approach at the heart of their hiring strategies will end up prospering most in the long-term.”

Glenn Elliott is the author of the 2018 bestseller, Build it: The Rebel Playbook for Employee Engagement.

Avatar photo

Aaron Hurst

Aaron Hurst is Information Age's senior reporter, providing news and features around the hottest trends across the tech industry.