Like it or not, artificial intelligence is changing how companies find, hire and manage talent.
AI for talent management is approaching peak hype. Hardly a day passes that a pitch doesn’t land in my inbox from a human resources technology startup touting a new AI service or an established vendor announcing an AI-based upgrade
Hype aside, it’s a fact that more companies are incorporating AI-based software into their people management practices.
It’s also a fact that while vendors believe the technology can make recruiting and other talent management functions more productive or fair, skeptics are just as certain it will cause more harm than good by perpetuating entrenched biases and discrimination. As examples, they point to computer algorithms some U.S. courts use for sentencing that have been shown to perpetuate racial stereotypes, and Microsoft’s experimental AI-based Twitter bot that turned racist and sexist within 24 hours of being launched.
Such concerns, though, haven’t soured companies on AI for people operations — far from it. In recruiting alone, close to two-thirds (62 percent) of 1,143 recruiters expect to spend more on AI-based software in 2018, according to a new survey from Entelo, the AI-based recruiting technology vendor. Of those, the vast majority (86 percent) plan to spend on AI-based software for sourcing.
As organizations invest more in AI, they’re also spending less on previous hiring sources. Almost 7 in 10 recruiters plan to spend less or nothing at all on traditional print help-wanted advertising next year, according to the Entelo report.
Early successes could explain the talent economy’s AI pick up. Hilton Hotels’ high-volume corporate recruiting division started using AI in prehire assessments in 2014 to fill call center and other customer support staff positions. Since then, the time it’s taken to progress from an initial interview to tendering an offer dropped from 42 days to five.
In addition to using AI for prehiring assessments, organizations are using AI to coordinate employees with projects that need their specific skills. This year, Royal Dutch Shell PLC tested machine-learning software that matched jobs on a project evaluating digital business models in the automotive industry with available employees with the desired expertise, according to The Wall Street Journal. The pilot project did well enough that in January the multinational energy producer plans to expand the Shell Opportunity Hub to 8,000 employees in its business-to-business marketing arm.
Organizations also are adopting AI to power chatbots that interact with job candidates or employees, approve requests for paid time off and improve wording used in job listings.
The Downsides of AI
Not everything that’s promoted as AI is truly AI, however. Some vendors have glommed onto the “artificial intelligence” moniker because it’s buzzy even though what they sell is more akin to data analytics, said Kieran Snyder, CEO and co-founder of Textio, at an HR Technology Conference panel in Las Vegas in October. The AI startup sells an augmented writing program to improve language recruiters use in job postings and other marketing materials.
Critics remain concerned that AI software meant to improve talent management processes could instead allow companies to perpetuate established biases in hiring and promotions. After all, AI doesn’t program itself, and whatever biases the employees or vendors setting up the programs have could show up in those systems.
A 2016 ProPublica investigation and academic research into analytics-based software that courts use to make recommendations about bail, sentencing and early release found that black defendants were twice as likely to be incorrectly labeled as higher risk than white defendants and received longer sentences as a result. At the same time, white defendants identified as low risk were more likely than blacks to end up being charged with new offenses.
Microsoft’s Twitter bot was programmed to learn about interacting with people by doing just that, but programmers didn’t account for the onslaught of racist and sexist remarks that led the bot to change its behavior, causing Microsoft to pull the plug on the experiment.
Skeptics and critics blackball AI because they don’t understand how it works, and because they think it will hurt jobs, said Kevin Parker, chief executive at HireVue, which sells the AI-based prehire assessment platform that Hilton Hotels uses.
According to Parker, HireVue’s prehiring assessment takes into account the potential for bias. The assessment analyzes performance-related criteria that led existing employees to be successful and matches those against job candidates taking the screening to determine if they possess the same qualities. The algorithm is programmed to factor out things like gender, age and ethnicity. For example, if all successful sales representatives had red hair, algorithms would be adjusted to neutralize that characteristic to ensure it wasn’t part of the criterion for success, Parker said.
HireVue customers have completed close to 1 million assessments since the company started selling the service four years ago. Though Parker says individual HireVue customers have become more diverse as a result of using AI, the company hasn’t polled its entire user base to see if the same holds true across the board.
Helping Jobs, Not Replacing Them
Instead of eliminating HR and recruiting jobs, AI ultimately could end up creating more than it replaces, said Andrew Chamberlain, chief economist at employer reviews website Glassdoor, at the HR Tech Conference in October. “It requires a big dataset to train on, everything needs to be labeled, and [it] breaks all the time,” Chamberlain said. “They don’t teach themselves, there are always people behind the scenes. People team up with it and end up doing more.”
For employees of all kinds, including recruiters and other talent management personnel, AI is a job transformer, automating things people are bad at such as high volume, repetitive tasks. “And it helps us do things we’re good at, like problem solving,” said Ray Wang, an HR tech consultant and founder of Constellation Research, also speaking at the HR Tech Conference.
Natural language processing — like the kind that digital assistants such as Apple’s Siri and Amazon’s Alexa run on — could eventually be used for job interviews and employee wellness and assistance programs, Wang said. It could also improve mentor programs by better matching mentees with mentors based on the experiences of each, and in efforts to reskill employees as companies grapple with rapidly changing technologies and market demands. “That’s where AI can help, figuring out how to get this individual from point A to point B,” Wang said.
Michelle V. Rafter is a business journalist in Portland, Oregon, reporting on workforce and technology for Talent Economy and other publications. If you have a comment or column idea, email firstname.lastname@example.org.