Humans Vs AI: Who Makes the Best Decisions?

It’s amazing what technology can do these days—sometimes even taking on roles and decisions that once required human consideration. Consider the potential of artificial intelligence, machine learning and predictive analytics, for instance, and the impact that these technologies could have on the field of human resources.

What if you could accurately predict which candidate would perform best in the position you have posted or in a management role?

What if you could accurately predict which employees are most at risk of leaving for another job?

Theoretically, you can already do these things and more using technology. But are the decisions that algorithms can make based on big data and predictive analytics necessarily any better than decisions seasoned managers or HR professionals might make based on their years of experience?

That’s a question that a recent Harvard Business Review case study explored. “Should an Algorithm Tell You Who to Promote?” shared the fictitious case of a hiring manager—the VP/sales and marketing for a global consumer products company—who was presented with data from the newly formed “people analytics team” that suggested a candidate other than her first choice would be the best pick for an open position. The study does not reveal which candidate was ultimately promoted, but both experts who weighed in on the story agreed that the final decision should lie with the hiring manager and her “human judgment,” not an algorithm.

We’re hearing a lot about the potential of machine learning, AI and other forms of emerging technology to disrupt the workplace as we know it. But, as the HBR case study suggests, machines don’t always know best. Predictive analytics can be used to identify employees who may be a good fit for a job, but not necessarily the best fit. In fact, as the hypothetical VP in this case points out, algorithms aren’t always entirely objective. After all, recommendations made are based on the data entered into the system—in this case subjective performance review information.

In short, intuition and experience can sometimes trump technology. But most of the time, say the experts, it’s some combination of technology and human consideration that best utilizes the insights that data and analytics now bring to the table.

AI Investment and Adoption

An Allegis Group survey of 7,000 employers published in November 2017, “Staying in Front: An Inside Look at the Changing Dynamics of Talent Acquisition,” explored some major trends, including the impact of technology. It may be heartening to hear that if your credit union hasn’t already embraced digital tools to help with various aspects of your HR processes, you’re not alone.

While respondents indicated that progress is being made to set the foundation for the use of technology like AI in human resources, and 23 percent of those responding indicated that they are investing in related innovation and R&D, only 13 percent indicated that they are currently using AI for talent acquisition and management.

Why such low adoption? There are likely a number of reasons ranging from “The technology is still changing and developing so we don’t want to jump in just yet,” to “We don’t have the budget to invest at this point,” and even, “We’re a little worried about the long-term impacts that AI might have on our credit union, our employees and our members.”

As such, most organizations are still in the very early stages of considering how this technology could benefit their HR practices, but many HR professionals say that the greatest impact will be in automating processes to free up their time to focus on more strategic considerations—like improving the effectiveness of hiring and promotion decisions.

Why Algorithms May Not Be ‘Right’

“Despite the intrigue that always seems to surround artificial intelligence and emerging tech, much of it is overhyped,” says Steve Wang, who has worked in HR for more than 15 years as a manager and recruiter. Wang has also helped build companies like Mock Interview and JazzHR from the ground up and is the author of a career blog. “Those who actually understand the algorithms and logic behind the predictive analytics would know that there are a lot of kinks and flaws to them,” he says.

Wang offers a familiar example—automated tracking systems that screen resumes using keywords. The systems, he says, look for keywords “which must be manually set by the employer,” and then rank resumes based on the number of keyword matches. But, he says, “Something this simplistic is obviously going to have flaws.” Good applicants may not make it through the system because they failed to use the identified keywords often enough. More tech-savvy applicants who are not as highly qualified may “keyword-stuff” their applications to allow their resumes to rise to the top. Both scenarios minimize the confidence that HR professionals and hiring managers might otherwise have with the technology.

Chris Hartman, global development officer with Allegis Group, Hanover, Md., agrees that there are potential shortcomings inherent in the use of technology for screening candidates and making hiring decisions.

“AI technology is a powerful tool for uncovering ... candidates who might have otherwise gone unnoticed, but HR professionals should not make their decisions based on AI insights alone,” says Hartman. “An algorithm isn’t an all-knowing entity. It was created by a human being, and its results depend upon the data on which it is trained. In other words, the creators’ biases as well as any biases present in the underlying data will likely show up in an AI’s findings.”

This is an important perspective to have. Too often, we view any data generated by technology as beyond reproach. Considering the inputs that drove the outputs you’re reviewing is an important step in critically evaluating their relevance and validity.

“Machine learning and artificial intelligence are designed ... to continually learn and improve with the assistance of data scientists,” says Adam Sbeta, cybersecurity analyst and senior team lead with Oakland Managed IT & Cyber Security Services, Oakland, Calif. But, he points out: “Machine learning in this space is fairly new and does not have the years of experience that an actual HR person has.” While machines can process data much faster and more accurately than humans, they should be viewed “as an assistance to HR and not a replacement,” Sbeta says.

Nigel Davies is the founder of Brighton, U.K.-based Claromentis, a company that offers integrated intranet, process management and employee engagement solutions. “If we had to distinguish people from machines in a word, it’d be empathy,” says Davies. “Computers have not been programmed to perform abstract thinking yet, which, for HR and hiring, is crucial.”

Add to the mix the subjective nature of most hires where qualitative factors like “fit” must be considered. “Skills and experience might be enough to get a person an interview or considered for a promotion, but organizations must never fall into the trap of box ticking,” warns Davies. “The personality fit has to be right for the culture of the credit union. It only takes one bad apple to upset the cart and change the dynamic of a well-functioning team for good with negativity or narcissism.”

Wang notes there's a reason hiring software is generally only used in the preliminary stages of the hiring process. “As it stands now, predictive models and software in HR are nowhere close to accurately accounting for things such as work ethic, employee potential, employee relationships, teamwork and human intention,” he says. For that, we need to bring the human element into the mix.

The Human Element

Experts in both technology and HR are united in believing that AI is not yet ready to overtake the human elements of decision-making related to various HR decisions—if it ever will be. It is, they say, a balance.

“Technology, and the data it can be programmed to capture, is a hugely valuable tool for fast decision-making or to bring HR to a sensible set of conclusions,” says Davies. “But these need to be put into context by a human.” In fact, more than one human, he adds. “Human decision-making is vulnerable to bias so, in the interest of fairness, more than one person’s intuition should be considered.”

Take accounting, a profession that has long relied on data and spreadsheets. Accountants don’t simply “crunch the numbers” and pass along the spreadsheets to clients. They evaluate what the numbers tell them—and consider the inputs that went into generating the final numbers.

“The experience and insight that an HR professional brings to the hiring process is absolutely critical to making good decisions,” says Hartman. While he believes that HR professionals should incorporate AI-based information into their decision-making processes when they can, “ultimately, a human being must make the decision. If experience and additional data sources lead the HR professional to make a decision that diverts from the direction the AI results are pointing, that is a very acceptable result and how the process should work.”  

Doomsday predictions that technology will one day replace the need for human beings are misguided, suggests Wang. “While technology will surely continue to evolve and play a larger role in the hiring process, I don’t foresee any revolutionary changes any time soon. Despite all the exciting talk about AI and machine learning, the algorithms and software used in most companies’ hiring processes today are nowhere near advanced enough to make our promotion or hiring decisions for us.” There are also cost considerations, he notes—the more advanced the technology, the bigger the price tag.

And—as we saw in the HBR case study—trust issues also come into play. “Employers and management usually pride themselves in their abilities to make the correct hiring decisions,” Wang says. Convincing them to heed what the data says about who to promote or hire is not going to be achieved overnight.

What AI Can Do Today for HR

Despite the uncertainty about how or whether technology will ever be able to replace human judgment in important HR-related decisions, there are ways in which credit union HR prof­essionals could leverage AI to save time and improve internal processes today, says Chad Davis of Kronos Inc., Lowell, Mass. Davis works in Kronos’ financial services industry team and has a background working with CUs. In fact, he says, employees are already seeing the potential benefits of the technology.

According to recent research from Kronos and Coleman Parkes Research, says Davis, “65 percent of financial services employees said they think AI could be used to simplify internal processes throughout the workplace—57 percent feel it could actually help balance their workloads.” He points to three specific applications:

• Smarter staffing. AI can be used to deliver more accurate forecasts to help managers meet staffing demands, says Davis. For instance, “If a credit union is using an appointment-setting solution with AI functionality on the back-end, managers can use that appointment data to match customer needs with the most-qualified available employees based on their skills and experience. Ensuring customers speak to the right employee the first time can significantly optimize service and sales.”
• Proactive labor compliance. Today’s technology, says Davis, can project up-to-the-minute timekeeping data, identifying and alerting managers to potential labor compliance risks hours or even days before an issue would naturally come to their attention. “This advanced warning allows managers to take swift action to prevent compliance violations from happening—thereby saving time, money and exposure,” he explains.
• Personal digital consultants. AI can be used to automate time-consuming daily decisions like handling employee time-off requests, says Davis, freeing up managers to spend more time on strategic initiatives. “Technology could be used ... to analyze variables like eligibility, accrual balances, staff availability, cost of replacement workers, and compliance with company policies and labor laws to rapidly make an informed recommendation,” says Davis.

While technology has historically replaced some jobs, in people-centric industries like financial services and activities like human resources, it is unlikely that technology will ever make humans obsolete. In the battle of AI versus humans, humans still have an edge.  

Lin Grensing-Pophal, SPHR, is a freelance writer and human resource management and marketing communication consultant in Chippewa Falls, Wis. She is the author of The Everything Guide to Customer Engagement (Adams Media, 2014) and Human Resource Essentials (SHRM, 2010).