Artificial intelligence may be making human resource departments less human.

“AI has the potential to automate some tasks in human resource (HR) departments, but it is unlikely to completely take over HR departments in the foreseeable future,” reads one ChatGPT response on the subject. “AI can assist with tasks such as resume screening, candidate matching, and even initial interviews, which can free up HR professionals to focus on higher-level tasks such as employee development and retention.”

This has long been the conventional wisdom on the subject. Many have argued that AI can replace repetitive job tasks, but not the human intuition that many roles require.

However, recent advancements in artificial intelligence (and the so-called Great Resignation) have precipitated a boom in, investment in, and adoption of products that pledge to radically disrupt HR departments—including those “higher-level” tasks that are supposedly reserved for humans, such as “employee development and retention.”

Companies such as IBM and Ultimate Software Group have paved the way for a new crop of “predictive attrition” companies such as Retrain AI, Eightfold AI, and HRSignal—that is, companies that inform employers when an employee is at risk of quitting.

HRSignal, for example, sells software that promises to predict if a worker is going to leave. The company recently raised $1.6 million in pre-seed fundraising from Gammite Ventures and investor Aaron Grossman. CEO Andrew Spott argues that his organization could revolutionize how HR departments develop and retain talent and that AI is set to eliminate entire jobs. “Over the next 10 years, you’re going to see some jobs start to get replaced by automation,” says Spott, who supports the idea of universal basic income. “In the long [run], AI is going to allow society to continue to function with less people. But I think in the short-term, there’s real potential for it to replace a bunch of jobs—jobs that probably people don’t like anyway—and it’s going to hurt families.”

Spott says that while it would have been easier to create a business that caters to recruiters, he and his cofounders were passionate about pursuing a mission to help workers get recognized and promoted. “We decided that the business model that we wanted to be in, and the way to use our algorithm responsibly, was to help people advance in their careers,” he says. “We’re hoping to be, as much as we can, an ethical data company.”

HRSignal buys wholesale data from data brokers, compiles various data sets together, interprets how likely a worker is to quit, and assigns them a risk score from one to 99. This analysis considers details such as a workers’ job title, where they live, the size of their organization, and the number of relevant competing job opportunities.

Spott says that HRSignal’s data set includes 400 million résumés that have been stripped of their names but does not take employer-collected data into consideration, nor variables such as age, gender, or race. However, his competitors do.

“We’re not alone in selling a product that tries to help employers retain more employees,” Spott acknowledges. He divides the sector into four categories: predictive analytics companies that use public big data to draw conclusions, analysts that interpret internal data provided by employers, surveyors that ask employees directly if they are considering quitting, and surveillance-focused organizations that employ spyware (such as keylogging and mouse movement tracking) to monitor if a worker is applying for new roles.

Across industries, employers “are getting smarter and smarter with predictive analytics,” says Frank Giampietro, chief wellbeing officer at EY Americas. Giampietro says that it is possible to identify “groups of workers that are more likely at risk [of quitting] than we would have realized in the past.”

And of course there are organizations who do not need to use AI to sense if an employee is at risk of quitting.

“We do something that doesn’t leverage AI,” says Marco Zappacosta, CEO of Thumbtack. “We look at our engagement surveys.” He says a traditional engagement survey has helped identify workers who are at risk of quitting and has also helped flag workers who are due for a promotion.

Humans have long been able to guess when a worker is considering leaving. Perhaps they are less engaged or less vocal in meetings. Perhaps they have taken an unusually high number of last-minute sick days or booked a sudden string of doctor’s appointments (classic covers for job interviews). Perhaps the pansophical “vibe” is off.

Using human intuition to flag workers who are at risk of quitting has worked for organizations in the past and continues to work for many today. However, the Great Resignation has shaken many managers’ confidence in their ability to retain employees, and many are now looking for evidence to support the decisions they were already planning to make.

Some have warned that using AI to make HR decisions could introduce—or reinforce—biases. Because of these fears, New York City is set to become the first U.S. municipality to regulate AI tools used in hiring. In 2022, 17 states introduced bills or resolutions to regulate AI.

Plus, predictive analytics often introduces new philosophical questions. For instance, what if an employee is only casually considering quitting, but after getting flagged by some form of AI as a quit risk, is now hounded by skeptical, or even angry, managers? What if these pressures cause a worker, who ultimately would have stayed if they had been left alone, to quit?

One of the most famous cultural examples of predictive analytics can be seen in the 2002 film Minority Report, starring Tom Cruise. In the movie, prophetic technology flags crimes before they are committed.

When asked if predictive attrition could be seen as a less-sinister relative of this hypothetical technology, Giampietro chuckles, and pushes back against comparing quitting to committing a crime. “Some level of turnover is good for individuals and good for organizations,” he says. “Quitting is ultimately a good thing.”