Job hunters are now using ChatGPT to respond to job ads that were themselves written by ChatGPT. It’s one way that AI is taking humans out of the loop in recruitment: AI-written job ads are scanned by another AI, which then writes a job application, which is then reviewed by another candidate-screening AI. 

“You can see how this gets silly pretty quickly,” says HR expert Nathan Wawruck.

Algorithms that match candidates to jobs compare text from the ad and a CV, explains Wawruck, director of permanent placement services at staffing firm Robert Half—but both are often now written by AI. “You’ve got robot-written documents being matched by robots,” he says. “And then people are sitting there wondering: why isn’t this working?” 

AI in hiring is the latest battle in a long-running war of attrition. Companies try to cut HR workloads by using algorithms to sift through CVs and cover letters. Applicants respond by using ChatGPT to respond to application questions and write keyword-hitting cover letters, while using other AI systems to auto-apply for dozens of jobs. Claims abound online of one bot letting users apply for thousands of jobs automatically.  

A November 2023 survey by Indeed found 92 per cent of Canadian recruiters were using AI in some way. Startups are looking to cash in. Calgary-based Job Autopilot claims it “outsmarts employers’ robots” and auto-applies for hundreds of jobs in just hours; Toronto-based Knockri uses AI to analyse short video responses from candidates; Kitchener-based startup Plum uses personality tests to match candidates with roles.

This back and forth between job seekers and HR isn’t new. Applicant tracking systems have long frustrated job seekers as they battle to ensure formatting mistakes or missing keywords don’t leave them on the “no thanks” pile. But AI is exacerbating the fight. 

The use of AI tools varies across companies and recruitment agencies. Robert Half, where Wawruck works, has in-house AI that matches profiles with roles, tasks that have long been done by keywords, filters and algorithms. “Sometimes you can use automated logic to speed that part of the process up,” he says. “More recently, some tools write job descriptions and summaries to help with documentation—that’s been helpful.” 

So far, Wawruck says, AI is largely being used to cut down on paperwork, freeing up HR staff to do more complex tasks. He says the tech is a long way from finding the perfect candidate with the tap of a button. 

But generative AI is also being used to communicate directly with candidates via chatbots, one-way video interviews and automatically generated emails. Lewis Curley, a partner at KPMG Canada with expertise in HR, said some firms are using chatbots to answer candidates’ questions about where an application is in the system and even to suss out and offer possible opportunities in the future. It’s the new “we’ll keep your resume on file”—instead, our AI will be in touch.

It can go wrong: recruitment is an industry with tight anti-discrimination regulations, and models are often trained on data that could make biases worse. 

In 2015, an Amazon trial AI system was scrapped after the engineers realized it had been trained on old data to methodically remove female candidates from the shortlist. LinkedIn’s early attempts to use job-matching algorithms favoured men over women by picking up on subtle differences, like the number of skills listed on a resume or how aggressively they engaged on the platform. And HR and finance platform Workday faces an ongoing class-action lawsuit in the U.S. after a Black job applicant claimed that its AI recruitment system turned him down for more than 100 jobs, saying a lack of guardrails made it “a ready mechanism for discrimination.”

Job applicants also aren’t on their best behaviour: a survey by Capterra of 250 Canadian job seekers revealed half of people are using AI tools to help with a job search, with most using it to “exaggerate or lie” about their skills on an application, while a quarter have used it to answer interview or skills-testing questions.

In an attempt to get a grip on the chaos, Ontario passed a law in March requiring companies to notify job seekers about the use of AI in the hiring process. The language of the law suggests the government is focusing on the use of AI in consequential decisions at the hiring stage, says Brian Wood, associate at law firm Norton Rose Fulbright Canada—but details about how the legislation will actually work have yet to be revealed.  

Given so many companies are using AI, simply knowing that fact may not be enough. New York’s AI governance regulations, known as Local Law 144, not only requires disclosure of the use of AI in employment decisions but details about how it’s used, including a bias audit. Ontario may look to that law to inform its own, says Wood. 

AI could shake up recruitment to improve it, rather than merely flooding the process with AI-authored slop. Plum CEO Caitlin MacGregor argues the current method of looking at a list of skills and past roles isn’t very effective—and AI trained on previous data amplifies the system’s flaws. “They’re just speeding up a broken situation,” says MacGregor. 

Plum’s personality testing is meant to match employees with roles they might not only be good at but actually enjoy, she says. Scotiabank used Plum’s system to revamp its university hiring scheme, ditching CVs and widening its hiring pool to more schools and subjects—for students, their limited past experience matters less than their potential.

AI could also help avoid discrimination. The Future Skills Centre and Toronto-based non-profit CivicAction built an AI tool called HireNext that assesses job postings for inclusivity, while also expanding the pool of viable candidates, according to Tricia Williams, FSC’s director of research, evaluation and knowledge mobilization. 

Rather than racing to hire, AI could help HR staff to slow down and spend more time on the human aspects, such as developing job descriptions or holding in-person interviews, while staying alert to their own biases. So far, though, it’s just accelerated the job-search war.