12 hours ago
Will a chatbot give me a job? I put AI recruitment to the test
The pot plant on the filing cabinet behind my interviewer's head needs water. Its curling leaves distract me as the floppy-haired recruiter in a crumpled black polo shirt poses his next question: 'In the role of tax consultant at McKinsey, balancing technical expertise with interpersonal skills enhances client relationships. Can you share an example of how your organisational skills and multitasking have benefited a specific task?'
I gabble through my recent work organising a large charity auction in my spare time, while juggling work deadlines and three busy children.
The interviewer nods, looking impressed. 'It's fantastic that you bring those soft skills to the table.' The grilling goes well until I hit some actual tax questions and stumble. My two decades in business journalism will not, it's clear, secure me a role as one of McKinsey's newest tax consultants.
But thousands of words of post-interview feedback arrive seconds later, and they are gently encouraging. 'You demonstrated your ability to mobilise resources and engage others … but need to practise emphasising how your unique abilities will benefit the company.'
Impressive — since the whole recruitment process was carried out by robots.
For £70 a month, website Final Round AI offers mock interviews with a human-like chatbot. Its questions are honed to any job listing you provide plus your CV. McKinsey had no involvement with the process, but one of its applicants might well have practised here first.
The CV I used was concocted by AIApply, which also makes cover letters specific to the demands of any job posting. A monthly subscription costs £23.
Before interviews, it provided a list of likely questions alongside what it deemed perfect answers. No need to learn them, though; for £29.99 a month the Interview Buddy app offers 'real-time assistance during job interviews, helping you answer questions confidently'.
Using this app during my mock video interview, it perfectly detects questions and rapidly bashes out uninspiring but detailed and accurate responses. Propping up my phone next to my webcam, I can easily reel off these AI responses to the, er, AI recruiter's questions:
This may sound outlandish, but it's happening: TikTok hosts thousands of videos of candidates undertaking real interviews while using ChatGPT-fed answers.
Human resources might need to be redefined as inhuman resources: hiring work often pits employers' AI systems against candidates' AI-generated responses.
Amid the frenzied talk about how AI is going to transform our working lives, in recruitment, it already has.
More than two-thirds of HR professionals use AI at work in the UK, according to research by tech firm The Access Group. That's far higher than in the EU, where it's only 38 per cent of firms, according to European tech firm SD Worx.
Anyone who hasn't applied for a job in five years will find being hired is an entirely new process. Applicant tracking systems (ATS) now scan and rank CVs and cover letters based on the inclusion of set keywords to whittle down the number of candidates, meaning a human may never see the majority of CVs sent in for a role.
Those who make it through may then face 'asynchronous video interviews', where candidates are filmed answering written questions that flash up like a PowerPoint presentation, before AI systems assess their answers, or chatbot-led interviews, like mine. Recruiters have the upper hand: the number of job vacancies in the UK has dropped from 1.3 million in 2022, to just 761,000 in the three months to April.
The concurrent adoption of AI by employers and applicants has created a race to the bottom. Firms are receiving an influx of thousands of chatbot-generated applications, while candidates report feeling dehumanised by the process.
'It was awful,' communications manager Laura Andrews said of a recent AI-led job application for a UK water utility. 'After an automated email saying I had been shortlisted, I had a video interview with a slideshow of questions listed on screen. It was really strange not having any kind of response from an interviewer to know how it was going. Without human interaction, my personality didn't show, and I think for neurodivergent people, it would be very hard.'
Weeks later, Andrews still hadn't heard back. 'I was ghosted, which is usual with roles now,' she added. 'I applied for more than 100 jobs in three months, and 80 per cent of the time I heard nothing at all.
'I couldn't contact the interviewer, as they didn't exist. When I chased with the recruiter, I was told it was a 'system error'. That's the flavour of the job search now: being AI-led, companies don't seem to care about candidates. The lack of feedback or even acknowledgement of applications that take hours to tailor felt isolating.'
This is a common complaint, according to Amelia Miller, whose start-up, Ivee, helps women return to work after a career break. 'Recruitment used to be a human process. Now, jobseekers are up against AI algorithms. The current market isn't sustainable. Recruitment teams are drowning in job applications with features like LinkedIn's 'easy apply' [which allows candidates to submit their CV in just a few clicks], along with hundreds of emails and messages. I believe this will lead to the death of the job application,' she said.
In the US, some recruiters have already moved to 'agentic AI', relying on bots to handle the admin of organising job interviews.
But while some claim that AI's transformation of the job search removes human bias, others say it embeds it. University of Washington research used real CVs but varied names associated with white and black jobseekers into AI recruitment systems, and found AI favoured the white-associated names 85 per cent of the time. Female-associated names were picked only 11 per cent of the time. 'If we're not careful, AI will just automate discrimination at scale,' Miller added.
Khyati Sundaram, chief executive of the ATS firm Applied, added: 'AI interviewers trained on datasets dominated by American speakers, for example, can be biased against candidates with different accents, non-native English speakers or individuals with disabilities that affect their speech.
'When assessing candidates' facial expressions, eye movement, hand gestures and tone of voice, biased bots can also favour those with a neurotypical communication style. Candidates who communicate differently can be ruled out of the running, regardless of [their] skills… reinforcing existing workforce inequalities.'
Not all candidates despise the changes, however. Rachel, a jobseeker who did not want her surname published, said ChatGPT and its rival, Claude, were 'huge timesavers.'
'I used them to refine cover letters, tailor them to job descriptions, and insert the right keywords,' she said. 'Recruiters are using AI, so I need to speak their language to pass the screenings. ChatGPT stripped out my personal attachment to my CV and showed me what was key and what was fluff, and what recruiters would ask me. It takes the legwork out of applications.'
To those returning to the job market after a long time, the recruitment landscape will 'feel very alien,' Miller acknowledged.
'The best thing you can do is learn how to use AI. Avoid applying for jobs on inundated sites such as LinkedIn and Indeed, and turn towards company websites and niche job boards to lessen the competition. Focus on building your network: ask for warm introductions and leverage the human element of recruitment as much as you can.'
The job market is being taken over by algorithms, but for now, the human touch still opens some doors.
'The market's more competitive than ever and the sheer number of tools and platforms can feel overwhelming,' says Kevin Fitzgerald, UK managing director of recruiter Employment Hero. 'Your CV has to be AI-literate.' Here are his tips on how to get it right:
Match the job description: Use the exact keywords and phrases from the job ad, especially for skills, tools and job titles.
Keep formatting simple: Stick to clean layouts. Avoid graphics, columns etc, as they can confuse Applicant Tracking Software. Stick to Word or PDF.
Use specific, hard skills: List tools or certifications relevant to the role such as Xero, Python, CRM management.
Quantify your achievements: Add real numbers where possible such as 'cut costs by 15 per cent' or 'trained five new hires'.
Avoid generic buzzwords: Terms such as 'motivated' or 'team player' don't help. Use language that's specific and meaningful.
Tailor your CV every time: Make slight edits for each application.
Use AI tools but edit carefully: ChatGPT and other AI tools can help draft or refine your CV, but always review for tone, accuracy and clarity. It should still sound like you.
