How Can Artificial Intelligence Work for HR?

Nationwide unemployment rates are low and technology can help employers find the best talent in a tight market—but HR can use artificial intelligence (AI) in a number of ways beyond hiring.

AI and automation could radically change the workplace and human resource management, said Alden Parker, an attorney with Fisher Phillips in Sacramento.

In the HR context, AI typically refers to data that is processed by algorithms to make decisions, he explained at the California State Council of the Society for Human Resource Management 2018 California State Legislative & HR Conference. "Machine learning can be used to constantly improve decision-making quality."

For the HR function, AI is most commonly used for talent acquisition. Forty-nine percent of respondents to law firm Littler's 2018 Annual Employer Survey said they use AI and advanced data analytics for recruiting and hiring. But AI isn't limited to talent acquisition. Survey respondents are also using big data to:

• Make strategic and employee management decisions (31 percent).
  
• Analyze workplace policies (24 percent).
  
• Automate certain tasks that were previously done by an employee (22 percent).
  

"It is encouraging to see employers starting to embrace the many benefits provided by big data in helping manage their most important asset, their people," said Aaron Crews, Littler's chief data analytics officer in Sacramento.

Litigation Strategy

Employment-related lawsuits tend to be fact-driven, which makes gathering documents and other information critical. However, only 5 percent of respondents to Littler's survey are using advanced analytics to guide their litigation strategy.

Employers may not be aware of the benefits to using analytics in this context, Crews said. "The ability to leverage data early in a case, to tease out insights before you ever take a deposition or begin evaluating the credibility of witnesses, is revolutionary."

Having the ability to find key documents lets the employer see what people were actually doing at a certain time and can help build a story, he noted.

Imagine that a repair technician who travels to residences to work on refrigerators files a wage and hour class action claiming that technicians weren't paid for all hours worked. The complaint and subsequent depositions will reveal the workers' version of what their days looked like, but analytics can be used to verify or refute their story.

"The more information you have, the better the decision-making process you can engage in," Crews said.

The employer could gather GPS data from work trucks, routing instructions, communications about the technicians' assignments, invoices, and cellphone and login information. These data will paint a picture.

The data may indicate that the technicians didn't work off the clock and were appropriately compensated for all hours—then the employer has some solid evidence on its side. But if the data reveals that the workers' claims have merit, it is better to review the data and know about it on the front end before going through lengthy and costly litigation, Crews said.

Pay Equity

Data analytics can also be used to assess pay equity. Legislation in this area is changing rapidly at the state and local level. For example, at least 12 jurisdictions have passed laws prohibiting employers from asking job applicants about their prior compensation. The idea behind such laws is to stop perpetuating historic discriminatory pay practices based on gender, race and ethnicity.

Technology can be particularly helpful to monitor employee compensation for discrepancies based on protected categories, Crews said, noting that some jurisdictions with pay equity laws have a safe harbor for employers that conduct audits and attempt to eliminate gaps.

[SHRM members-only toolkit: Managing Pay Equity]

There are tools available that make it easy to build a user-friendly experience and to analyze, understand and communicate data, he added. "You no longer have to rely on just an Excel spreadsheet full of math."

Getting a picture of what is actually happening in the organization is powerful, Crews noted. "When advanced technology is paired with good storytelling and visualization, it empowers HR professionals to have the conversation with the compensation team, executives and other decision-makers."

Chatbots

Certain technology, such as chatbots, can help employees access important information about policies and procedures from anywhere and at any time. Chatbots communicate by text and can be useful for answering common employee questions, Parker said.

Two-thirds of respondents said that they believe employees are more comfortable using chatbots than other forms of contact for transactional inquiries about paid-time-off policies, open enrollment and leaves of absence, according to a 2017 ServiceNow survey of 350 HR leaders. ServiceNow is a cloud computing company based in Santa Clara, Calif.

Employers that use chatbots need to ensure that they are complying with data security, disability and other federal and state employment laws.

Legal Pitfalls

When using AI to drive human resources strategy, HR professionals must monitor systems for bias. They need to look out for disparate impact—which happens when a seemingly fair or neutral standard is actually discriminatory in practice.

For example, a recruiting tool may weed out candidates that are more than 10 miles away from the worksite. What if the neighborhoods surrounding the worksite are predominantly made up of affluent white families? This hiring criteria could have a disparate impact based on race and ethnicity.

To reduce the legal risks, Parker said HR professionals should:

• Understand the legal theories that may be used to attack employers who leverage technology.
  
• Discuss potential legal pitfalls with technology vendors when evaluating products.
  
• Consider auditing systems for disparate impact, security and other legal issues.
  

"Bad input will for the most part lead to disparate impact," he said, noting that this issue is on government enforcement agencies' radar. However, if the system is designed to disregard irrelevant factors, it could reduce biases in hiring decisions.