Unbelievable! AI in Recruiting: Is it the CV Killer?

Students applying for graduate jobs this summer can take advantage of a new personal interview coach. If they submit a specific job description, they can receive personalized interview questions and answers – and feedback on their own answers – all for free.

The coach, offered by the Adzuna job search engine, is not human but an artificial intelligence robot known as a Prepper. It can generate interview questions for more than 1 million live roles at large companies, in industries ranging from technology and financial services to manufacturing and retail.

For a graduate job in PwC’s actuarial practice, the chatbot spits out questions like, “What skills do you think an actuarial consultant should have?” and “How would you explain actuarial concepts to a client who doesn’t come from a financial background?”. When a user answers a question, Prepper generates a score out of 100 and tells them which parts worked well and which parts were missing.

Prepper is part of a new wave of generative AI-powered chatbots, from ChatGPT to Bard and Claude. Chatbots are trained on large chunks of text taken from the internet, including books, newspapers, blogs, videos, and image captions. They can produce plausible, sophisticated text that is largely indistinguishable from human writing.

“In the last 12 to 18 months, he’s freaked out,” says Andrew Hunter, co-founder of Adzuna. “Obviously it’s hyped up a lot at the moment, but there are a lot of clever tools [to aid] recruitment and help people find work more easily.

AI is not a new tool for hiring and job hunting. Over the past decade, it has mainly been used to make processes more efficient and cost-effective for employers, from searching for keywords in CVs to filtering candidate video interviews.

But generative AI tools are rebalancing the power dynamic vis-à-vis candidates. “A lot of the recent improvements we’ve seen in AI are on the candidate’s side,” says Tomas Chamorro-Premuzic, an organizational psychologist and hiring technology expert. “A few years ago, renters were pretending to use AI to look cool even though they weren’t. Now they pretend not to use artificial intelligence.”

When Chamorro-Premuzic recently looked to hire for a role, he asked a candidate if they had any experience with generative AI. “They said, ‘If it weren’t for ChatGPT, I wouldn’t be sitting across from you right now.’” Their CV, cover letter and application had all been written by AI.

These systems are powerful, but they can also be wrong. How can we implement it with care and ethical attention?

Chamorro-Premuzic, who respected honesty and decided someone technologically savvy was worth hiring, hired the person. Others are less enthusiastic, warning that AI could signal the end of the traditional job application process.

“Generative AI can create great profiles – there may be some mistakes but only the individual will recognize them, not the employer,” says Matt Jones, of recruiting technology firm Cielo. “This raises the question of the relevance of reviewing CVs, cover letters and applications, particularly in the early career stage. I wonder if this is the death knell of the CV.

For graduates in an increasingly competitive job market, chatbots are a way to cope with a potentially overwhelming process. Ayushman Nath, a second-year student at Cambridge University, says many of his colleagues have toyed with ChatGPT, the public chatbot released by Microsoft-backed OpenAI, by asking him to write cover letters for specific companies. He knows people who have made it through early rounds or secured internships using cover letters and written questions from ChatGPT.

“From what I’ve experienced, he’s good at overcoming initial barriers. The first rounds of filtering are not personalized, they feel very remote and dehumanized. Everything is so automated,” Nath says of today’s recruiting processes.

Nath and his colleagues have also been subjected to automated video interviews run by recruiting technology providers like HireVue, which records candidates answering predetermined questions, usually with a time limit for each response. The recordings are sometimes looked at by the employer’s hiring managers; or the platform’s AI algorithms will evaluate each candidate’s performance, looking for various keywords from the job description.

The company hasn’t launched any generative AI products yet, but its chief data scientist, Lindsey Zuloaga, says her team is testing tools like chatbots for interview prep and new ways to glean insights from video interviews. “These systems are powerful, but they can also be wrong. How can we implement it with care and ethical attention? she says.

Grace Lordan, an economist at the London School of Economics and director of The Inclusion Initiative, which studies diversity in corporate environments, says companies, especially tech groups, are experimenting with generative AI to conduct initial interviews.

“One of the biggest areas of bias is actually the interview,” she says. “This is where people affinity bias, or representative bias, comes in. It means choosing people who look like others in the organization.”

AI-powered interviews might go some way to eliminating this bias, he says. “Generative AI is quite convincing as an avatar. Using AI as another serious data point will enable the pushback of machines [against human bias].”

More and more employers are using new assessment methods to broaden the pool of candidates they hire from, amid global skills and labor shortages and as they push to improve diversity. Automated systems designed to hire a more diverse workforce can find candidates who might otherwise be overlooked due to health issues, employment gaps, or because they lack a college degree or come from a non-traditional background.

But while ChatGPT is a useful starting point for a cover letter or for learning about a potential employer’s background, recruiters say it’s no substitute for writing an application yourself.

Nath, the Cambridge University student, says: “Companies are looking for a relationship with the people there, like a contact person in the company or a nugget of information that isn’t on the website. And these things can only be grown by personal interactions, not by AI models.”

Hunter, of Adzuna, agrees: “The caveat I’d give to job seekers is that AI can make a good copilot, but don’t let technology try to do everything for you. . . It’s a very nascent technology, it’s going to spit out cookie-cutter answers. If you let the initial interactions with the employer be completely handled by AI, then you won’t be able to get the job done.”