Job seekers vote in "Human recruiter v AI" contest (and it's not even close) - Ross Clennett: High Performance Recruitment Coach

0 Shares

In September 2024, I wrote about one of the newest and most successful startups in the recruitment ecosystem, Mercor.

Co-founders Brendan Foody, Adarsh Hiremath and Surya Midha (pictured, right), were all 21-year-old college dropouts at the time, and had just secured venture capital investment that valued their SaaS product that uses artificial intelligence to vet and interview job candidates and match them to open roles, at $250 million.

Fast forward thirteen months and Mercor has captured more headlines as they are reportedly on track to be the fastest company to reach USD $500 million in annual recurring revenue, having recently told investors it expects to reach that mark within weeks.

TechCrunch reports that if a Series C funding round is fully subscribed the value of the company will exceed USD $10 billion. It was only eight months ago that Mercor’s previous capital-raising round, a $100 million Series B, valued the company at $2 billion.

Unlike almost every other fast-growing startup Mercor is already in the black, generating $6 million in profit during the first half of the year, according to Forbes.

The company started as an AI-powered recruitment marketplace where applicants uploaded their resumes and completed a 20-minute video interview with Mercor’s AI. Approximately half of the interview covers questions about the candidate’s skills then the candidate has a case study to review from which the platform asks questions to assess the candidate’s knowledge and thinking. A second, job-specific AI interview might follow for a more specialised role.

Now Mercor reportedly provides contractors to Amazon, Alphabet (parent company of Google), Meta (parent company of Facebook), Microsoft, Nvidia (the most valuable company in the world), with OpenAI (owners of ChatGPT and reportedly Mercor’s largest customer by revenue). These contractors are domain experts who train and refine the foundational AI models being developed by the client company.

Last month, TechCrunch reported that Mercor was telling investors that it is adding more software infrastructure for reinforcement learning — a training method where AI model’s or AI agent’s decisions are verified or disputed, enabling it to incorporate feedback and improve over time.

As tech entrepreneur Swapnika Nag outlined on LinkedIn last month, Mercor has built one of the most important cogs for AI scaling:

AI models of today have already scraped nearly all the data on the public internet. There’s not a lot more to train on, and there’s a lot of noise to sift through.

For AI to become better at specialized tasks, models need to be taught by domain experts, e.g. to build a model that can accurately read a radiology scan, you need a training dataset and a test suite (or evals*) of labelled scans that the model can use to teach itself

Mercor has established itself as the go-between here (a GLG^ for AI) that matches experts with model companies to build this capability

*an eval refers to a structured test or benchmark used to measure an AI model’s performance on specific tasks

^GLG, or Gerson Lehrman Group, is an expert network that connects clients with experts to provide insights for decision-making. In the context of AI, GLG facilitates various applications and discussions related to artificial intelligence across different industries

In simple terms, Mercor is now a recruitment agency specialising in identifying and supplying the most valuable skill in the most valuable industry where there is the most at stake in the history of that industry (you might say, in any industry ever).

OpenAI CEO Sam Altman, said in mid-June that Meta had been offering its staff $100 million signing bonuses and “even more than that” in overall compensation in an attempt to poach them away (by the end of that same month eight Open AI employees had accepted Meta’s offer).

In case you need any additional evidence as to the amount of money on the table when it comes to identifying and hiring the right AI talent, consider the report, two months ago, that Meta CEO Mark Zuckerberg, offered Andrew Tulloch, an AI researcher and co-founder of Thinking Machines Lab, a compensation package worth up to USD $1.5 billion over at least six years (Tulloch declined).

It doesn’t take much imagination to contemplate how much margin potentially exists for Mercor in the world of matching talent with highly-paid opportunities when you consider the global staffing market produced around USD $650 billion in revenue last year.

Knowing Mercor’s product is already delivering sourcing, assessing and matching of candidates to jobs in the tech sector with a speed and scale unmatched in the traditional recruitment sector the potential profit available to Mercor, compared to traditional recruitment businesses still relying on significant human labour to do the same, is mind-boggling to contemplate.

It’s already clear from different user surveys across the past year that candidates are generally positive about the likely impact of, and happy with, hiring infused with AI.

Last year a survey of 3,000 UK and US workers reported that not only do 49% think AI could help the issue of bias and unfair treatment in hiring, but 46% of workers believe AI will be better than humans at being fair in the hiring process and 50% believe that AI could improve the hiring experience.

In the past two months three separate pieces of research into candidate preferences in hiring, when comparing human recruiters and AI recruiters, produced a comprehensive 3-0 scoreline in favour of the non-human experience.

Released last month, AI hiring platform Sapia.ai’s Humanising Hiring report draws on more than 1 million interviews conducted in more than 30 countries and 11 million words of candidate feedback about their experience of an AI-led hiring process.

The results were remarkable:

9.05/10 average candidate satisfaction across all groups and industries
81.8% of candidates left written feedback — engagement at this scale has never been seen before in hiring research
8 in 10 candidates would recommend an employer just because of the interview
30% more women apply when told AI will assess them, resulting in a 36% closure of the gender gap
98% hiring equity for people with disabilities through a blind, untimed, mobile-first interview design

Bullhorn’s GRID 2025 Talent Trends report, produced from a survey of nearly 2,800 current or recent customers of staffing firms located in North America, the UK and Ireland, Benelux, DACH, and APAC, found that of those who have interacted with AI in the job-search process, 77% rated it as positive with 88% of candidates rating AI voice agents either as good as, or better than, a human interview.

The report noted that dissatisfaction with communication during the interview process falls from 40% to 26% for candidates who interacted with an AI voice agent.

A study published in August by the University of Chicago’s Booth School of Business Center for Artificial Intelligence reported the results of job applicants’ behaviour in three scenarios when applying for roles in the healthcare, IT, and industrial sectors through PSG Global Solutions, a global RPO.

The study covered 70,884 applications for 48 different job postings located in the Philippines and 43 different client accounts (23 Fortune 500 and 20 leading European companies), predominantly for entry-level customer service positions.

The three scenarios were: Applicants interviewed by a human recruiter; applicants interviewed by an AI voice agent (“Anna”); and applicants given the choice.

The results were startling in that applicants interviewed by AI were 12% more likely to receive an offer, 18% more likely to start, 17% more likely to stay at least 30 days and, when given the choice, 78% of applicants preferred AI over a human recruiter.

Summarising the reasons identified by the various report authors in why a majority of respondents preferred AI over humans, it seems that AI is perceived as more responsive, timely, fair, and helpful (because of specific feedback about each candidate’s rating/performance compared to the selection criteria).

As is well known from years of research, “disrespecting my time” remains the major cause of job seeker resentment.

Although we are in the very early days of AI-led interviewing and assessment it’s clear that a significant proportion of job seekers have already cast their vote about what that disrespect means for the job search future they prefer.

Mercor shareholders must be very happy.

The stakes for AI-improved recruitment just got raised (by A LOT)
Recruiters remain terrible at assessing skills (from reading a resume)

GenAI takes temp recruitment back to the future

‘Disrespecting my time’ remains the #1 cause of job seeker resentment

0 Shares

Posted in Recruitment