On any given day, the number of muffins a Costco store sells can vary wildly. This variability made bakery staffing needs hard to predict. Schedule too few employees, and customers might not get the baked goods they want, but over hiring and overstaffing results in extra costs (and leftover muffins). It’s a problem common to many perishable products, although baked goods are especially prone to waste.

In search of a solution, starting in 2018 Costco tested using a machine learning model to predict the number of bakery products needed each day. Trained on historical trends, the model considered numerous factors, such as weather and local sporting events. Success followed: the 30-store pilot project reportedly resulted in nearly US$100 million in savings and has since been adopted company-wide.

Costco’s work highlights two underappreciated aspects of the use of artificial intelligence (AI) and machine learning (ML) in human resources (and, by extension, the operational work HR helps to supply).

New AI tools are impacting HR

First, AI is, of course, a lot more than just ChatGPT and its ilk. Conversations about generative AI dominate the media right now, but HR applications of machine learning predate generative AI and are often more advanced at this time.

Second, both AI and machine learning can be used in ways people have hardly considered yet. Even though its application is already widespread—the research firm IDC estimated that in 2023, 60% of the world’s largest businesses would use AI or machine learning “to support the entire employee life cycle experience”—and as the technology progresses, HR will use these tools to address an even wider range of tasks and challenges. These complex, iterative mathematical algorithms can recognize patterns and predict future developments, providing useful insights for HR professionals that touch on almost every aspect of work.

Hurdles remain: employee and candidate privacy requires protection. AI is susceptible to bias, depending on a raft of factors, from the underlying data sets—to the people using the tools. And without the right transparency, testing, and guardrails in place, many potential uses of AI will run into workforce resistance.

But the promised benefits make a compelling case for overcoming those hurdles. Here, we look at progress in current and potential uses in key areas of HR responsibility, from talent acquisition to training to succession planning, as well as how top companies are working to apply AI and machine learning in a trustworthy way.

Talent acquisition: scoring candidates, spotting adjacent skills

Talent acquisition is an exercise in pattern recognition, finding individuals with certain skills and interests and connecting them with jobs that match. This is especially important now, as technology is rapidly changing the work environment, making certain jobs obsolete while creating a demand for new roles that require new skills. This makes it a natural area to apply AI/ML. In fact, 2023 research by SAP SuccessFactors finds recruiting is the number one HR area organizations are currently investing in, with uses including:

• Résumé matching

• Automated scheduling and interviewing

• Interview-assisting chatbots

• Creation of more robust profiles for internal employees (helping match them to new positions)

Natural language processing has great potential for hiring tasks, looking forward, says Charles Handler, PhD, president of Rocket-Hire, a consultancy that advises large companies [on ethical talent assessment tools. AI can consolidate and analyze information from a range of inputs, like résumés, job descriptions, and LinkedIn profiles. Going further, Handler says, automated video interview analysis would involve breaking down spoken responses to help summarize and “score” interviews to assist human interviewers. Similarly, AI applications could analyze voice inflection in videos, phone calls, or in-person interviews, for traits like aptitude for customer service or sales.

However, these possibilities immediately surface the concerns about AI that HR must address. Hiring is particularly susceptible to AI bias. Any AI is only as good as the data it’s trained on, and various headlines in recent years—from an Amazon recruiting tool biased against women candidates in 2018, to a 2023 lawsuit alleging racial and age-related discrimination in Workday’s screening tools—can give pause to both candidates and employers.

Furthermore, employees are most comfortable with intelligent technology accessing work-related data sources, such as work calendars, active status, time tracking, and e-mails. They are least comfortable with technology accessing data related to their physical body, such as their tone of voice, eye movement, body language, or facial expressions.

To address these concerns, some states are introducing or considering laws related to the use of AI in hiring. For example, as Bloomberg Law reports, New York City’s Automated Employment Decision Tool (AEDT) law, effective July 2023, requires employers that are using AI and other machine learning technology in the hiring process to conduct annual third-party audits of their recruitment technology. Failure to comply with the AEDT law could result in fines of up to $1,500 per instance.

Simone Ahuja, founder of innovation strategy firm Blood Orange, stresses that HR leaders need to ensure that AI models are transparent and unbiased and—critically—that they are routinely audited. Candidates and employees alike need clear, consistent, and regular communication about how and why data is being captured, and how it will be used. The opportunity to offer feedback, express concerns, and see meaningful responses can help mitigate resistance and ensure AI is helping, not hurting the employer-employee relationship.

Learning and development: assessing skill gaps, democratizing coaching

One of the biggest challenges to learning and development is that the timeliness of training affects retention of the information. Like talent acquisition, just-in-time training also involves pattern recognition, but instead of matching people to jobs, it involves matching them to learning resources based on their current skills, interests, and immediate tasks.

In SAP SuccessFactors’ research, overall learning was the second highest area of organizational investment. Within this area, the most common application was in identifying recommendations and pathways for what employees need now, what they will need next, and what skills organizations need. Generative AI is also being used in virtual and augmented reality, allowing employees to have immersive experiences that can approximate challenging or dangerous real-life situations, such as in training for firefighters or surgeons.

The idea of combining AI with game frameworks presents interesting new possibilities, according to Tim Dasey, author of the 2023 book, Wisdom Factories: AI, Games, and the Education of a Modern Worker. “The tech is already able to do this, and some companies are incorporating AI-infused games in HR-related products, but it hasn’t become widespread,” he says.

For instance, companies can present employees with different scenarios to review their work output—useful for assessing their management, leadership, and interpersonal skills. Dasey offers an example: “A candidate could be asked to reach an agreement on an issue with an avatar that has been given certain prominent personality characteristics relevant to the job, so that the candidate's ability to adapt an approach to a personality can be assessed.”

“Or a leader could be put into a tradeoff situation where they must balance something like productivity with worker engagement or satisfaction, with an AI evaluating everything from how they manage resources to how they communicate with workers,” he says.

AI’s scalable automation can help “democratize coaching,” Handler points out, making it accessible to employees at all levels—not just senior leaders or employees identified as having high potential.

AI could also help to identify which employees serve as de facto trainers or coaches. For instance, Genpact (part of GE Capital) is a professional services firm that manages business processes for companies around the world. Genpact uses an AI tool to find “knowledge nodes” within the company, identifying go-to employees that others tend to turn to, for their assistance in creating training materials on their areas of expertise. They’ve identified about 700 internal experts who are contributing their knowledge to courses on more than 600 skills.

At Cigna, many development tasks are framed in terms of career path planning for its workers. The company uses an AI-powered tool—similar to Netflix’s recommendation engine—to collect data on employees’ strengths, skills, role duration, and growth aspirations, and then suggest opportunities aligned with their career stage, interests, and qualifications, according to an HR Brew interview with Amanda Day, Cigna’s VP of talent acquisition.

Career path and succession planning could benefit from more advanced applications for aligning people skills and talent potential to future needs—for example, using machine learning to identify existing employee skills that could be applied in different positions, along with the training needed to fill those new roles. Customer service skills may be applicable to business development or sales work; employees in finance or programming may be able to shift into data science jobs. This could allow companies to be more expansive in how they recruit, and in helping employees discover career paths they might not have considered but which they have the competencies to excel in.

Talent mobility: matching workers with projects

As the pace of change increases, companies need to improve their ability to realign their workforce to meet shifting business demands. AI can analyze employee assessments and other types of employee data to help identify workers for project teams or open positions of higher value. These applications could also be useful in identifying gig or contract workers to augment an organization's internal skills.

While internal mobility was the fourth highest area of AI investment in SAP’s research, it’s also an area that organizational leaders have identified as a higher priority for the future.

Talent or opportunity marketplaces[DS1] are one way this is done, as MIT Sloan Management Review and Deloitte note in a Future of the Workplace Report. Talent marketplaces match individuals to opportunities based on their skills and interests. Pointing out that while 74% of respondents believe employee development is important, only 34% are happy with their investments in this area, the report’s authors Michael Schrage, Jeff Schwartz, David Kiron, Robin Jones, and Natasha Buckley conclude that strategic investments in digital tools, leadership, and culture can transform opportunity marketplaces into sophisticated, adaptable systems.

Engagement and retention: analyzing sentiment and spotting attrition risk

In our research, increasing retention was one of the top reasons HR leaders indicated that investing in intelligent technologies was important.

Employee engagement is crucial for organizations today to help them effectively use the talent they have in-house to boost productivity and reduce turnover. Leaders and managers, as well as HR pros, can apply AI and ML in a variety of ways to boost engagement and retention.

For instance, predictive analytics can identify attrition risk. AI can help tailor and personalize training programs to address specific employee needs and interests for personal and professional development. ML can analyze workloads and task assignments to minimize the potential for burnout.

Furthermore, sentiment analysis software can analyze e-mail and other communications to help identify if leaders and managers are giving enough, appropriate, supportive, and culturally aligned feedback to employees. Genpact, for instance, uses a chatbot to measure employee mood and sentiment, asking employees questions at least four times a year, often based on prior interactions. Genpact’s HR chief, Piyush Mehta, says this process is faster and gathers better data than “regular” employee surveys, according to coverage in Yahoo Finance. Furthermore, in 2020 the company tied 10% of its CEO and top leaders’ bonuses to these mood scores.

Organizational design and staffing: improving precision, analyzing information flows

Workforce planning tools powered by AI and ML can be used to match current workforce skills with future workplace needs. We saw an example of this in how Costco is using technology to project product demand to help better align staff with desired productivity. Operational leaders are likewise applying AI/ML to understand work processes and flows. Putting these streams together, in the future, AI may help refine the overall structure of an entire organization.

In healthcare settings, AI could be used to help predict the need for clinicians and allied staff to serve patient needs based on predictive analytics applied to covered populations and available beds, for instance. In large organizations of all kinds, scheduling can be extremely complex. Self-scheduling technology can be used to help manage schedules while also allowing flexibility for employees.

In the future AI may help refine the overall structure of an entire organization.

Staffing issues related to safety can also be addressed through AI-driven technology. For instance, telecom provider NTT used Internet of Things (IoT) technology along with predictive analytics to identify drivers potentially at risk of becoming inattentive due to fatigue after noting a rise in deadly transportation accidents.

Job design can also be approached in creative ways by integrating AI into work processes. For instance, radiologists working with AI digital assistants to streamline repetitive processes.

At its broadest application, AI may help companies better understand how all these processes and employees interact. For example, Organizational Network Analysis (ONA), a formal discipline for studying decision-making and information flows, originated in the 1980s but can be supercharged by today’s exponentially more powerful data capture and computing power.

Working together for smart AI use

Handler points out that AI is a tool, not a solution in itself, and stresses the importance of understanding the nuanced applications and ethical considerations involved. He acknowledges that AI is underutilized in HR and urges HR professionals to experiment with AI, even on a small scale, with “tiny projects that align with their business goals.” This approach, she says, allows HR leaders to “gain familiarity with AI’s potential and benefits without extensive risk.” Ultimately, he says, AI has the potential to amplify, rather than replace HR’s work.

It’s not a journey that HR leaders can, or should, navigate on their own.

The development of organization-wide policies on appropriate AI use will require collaboration with business, legal, customer, employee, and human resources professionals, Dasey says. “Since AI is moving so quickly and any initial policy determinations are likely to need revision as experience is gained, there should be a standing group that accumulates expertise on the matter and adapts policies and norms on a regular basis.”

Genpact illustrates the importance of this approach. “Twenty to 25% of our efforts don’t go into project execution, but they go into speaking with legal and (information security) teams,” Praful Tickoo, Genpact’s vice president of data, insights, and AI, told Yahoo Finance.

The future of AI in HR, Ahuja says, will be a blend of advanced technology with a strong human touch. AI can never take the place of humans when it comes to tasks that involve the concept of humans caring for other humans which, of course, is what human resources is all about. AI can, though, free up HR professionals’ time to allow them more time to focus on areas that require their uniquely human touch.