People Analytics and Artificial Intelligence: Opportunities and Pitfalls
Edgeworth Economics President Chuck Fields and Principal Consultant Dr. Stephanie Cheng recently presented on this topic at the DC SHRM 2022 Annual Conference. This post summarizes their key takeaways from that event.
Artificial Intelligence (AI) is a rapidly growing field in computer science with a wide range of applications that, increasingly, includes the field of human resources. Many AI applications are now available for key HR processes like recruiting, retention, work allocation, and even employee engagement. While these relatively new AI tools offer powerful new ways for HR leaders to add value to their organizations, they also come with new challenges. If an organization is not thoughtful when introducing these tools, the results can often be counterproductive or even lead to potential legal risks.
Defining and understanding what exactly is meant by “artificial intelligence” can be confusing due to the ongoing rapid evolution of the field. Generally, we can think of AI as the creation of intelligent machines that can act and make decisions like humans. One commonly used subfield of AI is machine learning (ML), which focuses on how machines process, interpret, and adapt to data without intervention by humans. With ML, a computer is fed a “training” dataset—information like applicant data, employee data, or text such as emails and chat messages, and even photos. The model then algorithmically looks for patterns in the data and makes predictions. Once a ML model has been developed through this training process, the program can then apply the model to new data and predict outcomes.
HR professionals who are already familiar with AI/ML technologies were likely first introduced to these tools in their recruiting processes. Many large companies are now regularly using AI/ML to more efficiently sift through their applicant pools and identify candidates most likely to succeed in their organizations—often by finding applicants with profiles most like their current high-performing employees. In fact, research has shown that AI/ML techniques can identify employees that stay longer and perform as well or better than employees selected by experienced managers. Similarly, AI/ML techniques have been used for a number of years to assess employee retention risks, alerting HR leaders of employees who are at the highest risk of leaving and allowing those leaders to better respond and improve retention rates.
But as AI/ML tools continue to evolve, HR leaders have started finding new ways to apply this technology to improve existing processes. For example, by harnessing AI/ML to perform real-time analysis of a company’s email, messaging, and video conferencing platforms, HR leaders are gaining new and faster insights into employee engagement and social connections within their organizations that weren’t available previously. These tools also provide employees valuable information that help them better manage their own work-life balance. Further, AI/ML technologies are being used in other unexpected ways. For example, natural language processing—a form of AI focused on teaching computers to understand written and spoken language in a similar manner as humans—is being used to create chatbots that can converse with employees or job applicants and answer frequently asked questions in a natural way. This innovation provides candidates and employees with effective answers while providing recruiters and other HR professionals with additional bandwidth. Recently, AI technologies have even been in the news as a tool used to select individuals for reductions-in-force.
AI/ML applications can benefit the HR function in a number of ways, such as by identifying unexpected trends and automating time-consuming tasks. Unlike many humans, AI is well-equipped to handle large, complex datasets—in fact, its performance often improves as it is fed more data. AI also acts consistently over time and thus can automate overly repetitive tasks that would otherwise be time-consuming or potentially prone to human error. This leaves more time for HR professionals to handle strategic tasks or specialized one-time projects that could not be automated.
At the same time, it is important to understand the potential pitfalls of using AI before implementing these technologies. For example, companies implementing these tools may have to carefully manage trust issues with their employees, as new users may be skeptical or wary interacting with AI. Similarly, depending on the data used, organizations may have to navigate state or national data privacy laws. And like any other tool, AI is only as good as its implementation. AI/ML techniques also require large enough datasets to perform the “training” step described earlier. For small and mid-size companies that only employ or hire a relatively small number of employees, AI tools may not currently be practical or cost-effective. AI can also be prone to errors based on the training data provided. For example, a popular AI text-to-image generator recently created a number of humorous (and completely fictitious) candy bars when queried by users for images of popular Halloween candy. On a related but more serious note, AI used for HR practices like hiring, pay, or termination decisions can also be prone to the same biases as humans, which could lead to potentially discriminatory outcomes (and lawsuits).
Given the continuing evolution in technology and expansion of available data, the future of AI likely holds even more applications than what we see today. It is up to the human users of these tools to be cautious of the pitfalls described above and to implement AI into their systems thoughtfully and effectively. When successfully implemented, these AI-human partnerships have the potential to improve organizational effectiveness and enhance the employee experience.
 See, for example, https://joshbersin.com/2021/02/the-massive-market-impact-of-microsoft-viva/.