How artificial intelligence helps companies recruit talented staff

Written by: Emma Jacobs
Published on: 7 May 2019

Source: Financial Times, published on 25 February 2019

When Pitney Bowes, a business services company, recently opened an ecommerce fulfilment centre with the capacity to process up to 44,000 parcels per hour, it also had to substantially increase the workforce. To do so, it used artificial intelligence. 

It works as follows: after candidates arrive at the company’s careers webpage, they are greeted by a chatbot which shows them positions opening in their area and then takes them through the pre-selection process. This includes questions such as, “Are you able to lift 50 pounds?” “Do you prefer working a day shift or are you flexible?” 

It can then schedule an interview and send an email invitation. Brigitte Van Den Houte, vice-president, global talent management at Pitney Bowes, describes the chatbot’s tone as “welcoming”.

She says: “For areas where we have a high-volume recruitment challenge, AI has helped us to scale efficiently without having to temporarily deploy additional recruiters.” The use of such tech, the company says, has helped reduce the time to fill some positions by 10 per cent — or nine days.

A growing number of human resources departments are deploying these technological tools to find recruits. A host of recruitment tech start-ups have sprung up to meet this need. Barry Flack, an HR tech consultant, says that these have been attractive to investors as they promise to solve a problem they “have all experienced”. 

New ventures include some that search online and in databases for matching candidates, sifting through vast quantities of applications and using machine learning. “Finding matched profiles has become easier,” says Mr Flack.

Then there are chatbots that can answer candidates’ rudimentary questions and help screen applications at the first stages (for example, basic technical and experience requirements and legalities such as the right to work). Further down the recruitment funnel, there are online assessments using game theory and psychological profiling. 

Some employers are deploying video interviews and others use basic AI to help discern if candidates are “confident” or “passionate”.

Candidate relationship management tools have also grown in recent years. These help employers market their brand to prospects who might be interested in applying and nurture potential candidates with campaigns and communications, hoping to convert them into applicants.

Such tools have shaken up recruitment. A working article entitled “Artificial Intelligence in Human Resources Management” argues that the “speed with which the business rhetoric in management moved from big data to machine learning to artificial intelligence is staggering. The match between the rhetoric and reality is a different matter, however.” 

One of the biggest HR tech claims is that technology can sift through applicants to find the best fit, free of human prejudices. Alistair Cox, chief executive of Hays, the recruiter, wrote in a blog post: “as well as increasing efficiency, automating sections of the screening phase can also lead to a decrease in subconscious hiring bias”. 

Peter Cappelli, director of the centre for human resources at the Wharton School, University of Pennsylvania, and one of the authors of the article, is sceptical. “Any kind of structure eliminates bias. If you told employers to standardise the questions they ask that [would] eliminate the bias. A lot of [tech] imposes structure on the hiring process.”

As the oft-cited example of Amazon shows, technology can be as biased as humans if it replicates past hiring decisions. The tech group had to abandon its AI recruitment tool when it realised it discriminated against women because it tried to find candidates much like its current workforce: in other words, men. Decisions made by algorithm may also make a company more vulnerable to legal action, argues Prof Capelli, because the assumptions driving hiring decisions are clearly set out.

Technology can also require considerable finessing. Deloitte’s Human Capital report described one tech vendor as having “taken over a year to train its chatbot to intelligently screen hourly job candidates”.

Andrew Chamberlain, chief economist at Glassdoor, the jobs review site, says that automating trawls through the internet, looking for recruits on LinkedIn and other online platforms, has drawbacks. “Not all candidates show their best side [online]. They get left out of searches.”

Some candidates, by contrast, are brilliant at it, says David D’Souza, membership director at CIPD, the human resources professional body, citing LinkedIn profiles: “Everyone is ‘strategic’, everyone is ‘good with people’ and everyone is ‘proactive’. At the point that everyone becomes all those things you lose the ability to differentiate.”

Other products are focused on building a pipeline of interested candidates — including those that have failed previous applications and ones who register interest on an employer’s website. It will then market the company to candidates, and build a relationship. One such tool is Beamery, which calls itself a talent engagement platform. It gauges a candidate’s interest in the company or job (for example, if they open an email from the company) to determine whether they are ready to pursue a role. 

Products that allow a recruiter to keep in touch with failed — or potential — candidates can be useful. However, Prof Cappelli says: “Don’t kid yourself that this is a relationship, that the company really cares about you.” Moreover, the risk is that they bombard people with marketing material. “If you are the first mover it’s a smart thing to do but not the 20th.”

“One of the things that people have always valued is human connection, whether with a boss, or clear feedback at the end of a process,” says Mr D’Souza. “Part of the challenge with tech is that it’s more efficient but it reduces human contact. People want speed and convenience, ideally from a person.”

The candidate’s experience is critical, agrees Bill Boorman, who advises recruitment tech start-ups. They want to understand the process and receive feedback. “They don’t want it to be painful.” 

Ultimately, as with automation of other white-collar jobs, the hope is that technology reduces repetitive tasks. Matt Weston, UK managing director of Robert Half, the recruiter, says: “It’s really important to get the balance with the human technology. At the end of the day HR is a personal department that can’t be too automated.” 

Prof Cappelli says that in an age of techno-hype, there is an expectation that machines will solve the recruitment problem. “In a typical company you can persuade the CFO to buy software. Does it work? Probably not.”

The technology might even be making things worse. When companies use a new tool, says Mr Chamberlain, they tend to overuse it. New technology makes it easy to add extra tests.

“The time for the interview process has got much longer . . . They are making people jump through more hoops,” he says. 

“It’s not clear that it’s improving the match of candidates. It will take trial and error to see what tech really delivers on these promises.”

Recruitment tech to help weed out bad hires

Some recruitment tech is working with psychologists to devise psychological quizzes to discover the motivations or personality types of a candidate or employee. 

Attuned is one such company that offers its product for both recruitment and retention. It asks the person taking a test to answer a series of questions, to find out if they are motivated by things such as money or a desire for autonomy, to see if they are a good fit with their team or company.

Casey Wahl, the founder, says that one recruitment company that used it for hiring found that they reduced the number of bad hires — so that instead of 30 per cent of recruits leaving within six months, it was only 10 per cent.

Copyright The Financial Times Limited 2019

© 2019 The Financial Times Ltd. All rights reserved. Please do not copy and paste FT articles and redistribute by email or post to the web.