Companies are now recruiting artificial intelligence to help with hiring — but bias issues still exist
If I were the kind of person prone to telling tall tales, I would say that my career has been a product of hard work, talent and ambition. In truth, though, it’s mostly been luck — and knowing the right people.
The reason you are reading these words right now is not because I am the second coming of Proust, but because I said hello to an editor at a Christmas party in the late 2000s.
Is that not the way these things go? For all the talk about education, skilling up and more, countless Canadians find their positions through networks. In so many cases, it’s a referral rather than resume that matters.
Enter technology. In the digital era, technology is supposed to be the great arbiter — the way in which we supersede mere human frailty and give everyone a fair shot.
It’s not just an idea. Artificial intelligence and HR company HireVue recently conducted a survey of 6,000 talent managers in the U.S., U.K. and Australia and found that 30 per cent were increasing their tech budget, mostly reflecting a growing comfort with the use of AI in the hiring process.
Whether or not the use of AI is a step forward in hiring is a matter of some debate though. AI might have its uses, but it can also introduce or reinforce bias into an already problematic situation.
Part of what is driving a need for some kind of digital solution is technology itself. While it was obviously no picnic sorting through paper resumes, now that job applications are almost all done electronically, even quite ordinary positions can now garner hundreds or even thousands of applications.
That same HireVue survey suggested that around half of their respondents were reacting to such a deluge by automating the process. Not only does it allow a company to process many applicants, it also frees up recruiters to do more important tasks.
It might be worth noting at this point that HireVue is a company that pitches its automated and video solutions as a way to streamline and improve hiring.
That is the pitch for the use of AI in hiring, after all: it will do it better and faster. But that also comes with downsides.
For one, AI isn’t a mind doing things; it’s a pattern recognition machine, and those patterns come from pre-existing data that it is trained on. A 2023 academic paper in the journal Humanities and Social Sciences Communications stated that, despite some possible benefits to use of AI-like technology, algorithmic bias results in discriminatory hiring practices based on gender, race, color, and personality traits.
An example might be an AI filtering system trained on the resumes or skills of software engineers, an overwhelmingly male field. Such systems then have a tendency of down-ranking CVs from women, while elevating those from men.
The trouble with automating processes is that in order for it to work it has to account for pre-existing biases that exist in societies at large. This is why AIs aren’t actually intelligent. When people talk about AI today, what they are most frequently referring to are in fact algorithms, software that filters or chooses.
That choosing is not some mythical objective act, though; it’s a repetition of existent prejudice.
That’s also why, say, face recognition or photography software has historically had trouble with darker skin, or why algorithms designed to help establish credit scores can in fact marginalize lower income or radicalized peoples.
The use of AI in hiring is thus a fraught endeavor. In outsourcing part of the process to technology, one dilutes the unexpectedness or idiosyncrasy of the human.
One stark example is the artist Anthea Mairoudhiou, who was temporarily laid off during the pandemic and asked to reapply. The AI software who evaluated her spit out a result that her body language was marked poorly. She lost her job, but later settled out of court after taking legal action.
The maker of the AI evaluation software? HireVue.
All the same, there is a kind of intuitive logic to the idea of using tech in hiring. Human beings are fickle, emotional, biased. That we want something else to overtake subjective decisions is understandable.
There are some attempts to use AI to do the opposite and counteract bias. Toronto-based company Knockri attempts to use AI to help companies hire diversely. At least according to its own figures, it has helped clients boost diverse hiring by 25 per cent.
Yet, the automation of hiring with what we call AI seems far from a panacea. To the contrary, without careful, deliberate intervention, it appears to merely embed or reproduce the deep injustices of the past.
Getting a job via networking was never meritocratic or even fair. In trying to get past that less than ideal situation though, the risk is always in losing the human touch.
