The experts weigh in: what role does artificial intelligence play in the recruitment industry?
Mentions of artificial intelligence often conjures up Hollywood-driven fantasies of robots attacking humans or all-knowing computer sidekicks, but the reality is AI is already helping professionals in the recruitment and staffing industry place candidates in new jobs.
The ethics and pitfalls of AI in recruitment is a hot topic among RCSA members – prompting CEO Charles Cameron to turn to the experts - CuriousThing Co-Founder David McKeague and Compono Co-Founder Rudy Crous – to discuss the technology in a recent webinar. David is an artificial intelligence entrepreneur with 20 years’ experience building tech-powered solutions to support industries to work more effectively. His state-of-the-art AI product CuriousThing was designed to explore questions more intelligently for deeper insights that are measurable.
As a corporate psychologist and organisational culture specialist, Rudy has worked extensively with businesses to improve their people systems and create human-centred environments. His industry intelligence solution, Compono, is designed to optimise employee engagement across three key HR pillars including talent acquisition, retention and development.
While both experts were quick to dismiss an immediate war with robotic overlords, the say the truth is we’re already using AI every single day.
“There is a lot of fear mongering about machines taking over the world, becoming conscious and making decisions - generally, we’re nowhere near that. What we have is augmented intelligence and technology that gives us humans the time back to focus on human skills – emotional intelligence, engaging and interacting with each other,” Rudy says.
David adds that we are already using “narrow AI”, which he defines as mathematical and statistics driven data techniques, several times each day.
“When we think of a Google search, when we think of advertising; it’s an algorithm that determines what we see. The Netflix recommendation engine; that’s another piece of AI. Facebook and the notion of self-driving cars – that’s also AI,” David says.
“However, there is a distinction between an algorithm and AI. To be AI, the algorithm has to be learning and improving. So, if we have a chatbot that only ever has a conversation in the same way – that would be an algorithm. But if the chatbot learns, evolves and improves the conversation over time – that’s AI.”
For recruiters, AI systems can help match the right candidate with the right role. Currently, the majority of hiring decisions are still based on what is written in a CV, particularly a candidate’s skills and qualifications. Rudy points out this is no guarantee a candidate will be a good fit for an organisation – or that they will perform well in a job.
“How good are we humans at making hiring decisions? I’d argue that we are terrible,” he says.
“Because if you look at job satisfaction, most people are actively disengaged or not engaged in their jobs, and a large proportion of HR managers wouldn’t rehire someone they’ve employed if they had the choice.”
Rudy adds that organisations and hiring managers rely too heavily on the unstructured interview – or the ‘would I like to have a beer with them’ test.
“An interview, like the CV, has very poor predictive validity. Us humans, we kind of fool ourselves into thinking that we are good at decision making – but we’re really not. This is where technology can really help to standardise our approaches to screening, ranking and matching workers.”
AI can help recruiters avoid unconscious and inherent bias by delaying human judgement in the hiring process. AI can help screen candidates by assessing skills and qualifications as well as determining how well an individual fits into an organisation. It can speed up the recruitment process, and limit unconscious bias. But badly set up software can create its own problems.
“If you just throw machine learning into data sets, you can get some really bad predictors,” Rudy explains.
“For instance, there is an overrepresentation of males over females in software development. If you were to throw statistical analysis or a data tool into this, it would look at the data and say there is such a large proportion of males, so maybe it is gender that is a key predictor of performance.
“What this data has done has magnified a noise that really doesn’t predict anything. This is where we, as technologists, need to provide freedom within a framework for the data. Our matching algorithm strips out the noise factors like age, gender, name and what university they attended to strip out the inherent biases.”
Both experts stressed the importance of recruiters understanding how the technology works, before investing in AI recruitment software. If vendors are not forthcoming about how they designed the program, how it works, the academic model and the validity – that’s a red flag. Recruiters also need to understand the ethical implications of AI – which should be fair, transparent, safe, reliable, and respects candidates’ privacy. Finally, they need to appreciate the limitations of robotic recruiters.
“As a recruiter you have to look at the data – and everything else as well,” Rudy says. “It is up to you as the person using the software to understand how it works and will make you better at your job. AI technology should never be used to make the final decision – it should guide the decision.”
Questions to ask before investing in AI technology
Is the assessment reliable? Does it actually provide consistent results?
How good is the predictive validity? If someone scores high, how do I know they are going to be good at the job?
What is the construct validity? What is the academic framework the underpins the algorithm?
What is the concurrent validity? How is my assessment that is measuring culture or personality relate to other measures of personality? Is there a correlation between the two?
To watch the full webinar, click here.
The password is: Q8me?&RT
