Introduction to Special Issue: People Analytics

Scholars and practitioners operating within organizational and institutional contexts should heighten their focus on the innovation, adoption, and impact of technologies, particularly artificial intelligence (AI) and data analytics. AI-based solutions, especially generative and multi-modal foundation models, are rapidly reshaping how people work, learn, and develop, a trend that will only intensify in the future.

To illustrate, the recent introduction of ChatGPT has spurred an unprecedented adoption of AI, fundamentally altering how individuals search for, generate, and consume information. While AI and data analytics are closely intertwined, the latter often doesn’t receive due attention. Before delving into people analytics, I will clarify its meaning, importance, and role in the broader landscape of HRD research on AI and technologies. This special issue comprises articles with titles that include both terms, and some may wonder whether People Analytics (PA) is AI or what roles AI plays in PA, necessitating brief clarifications.

Analytics lies at the core of applying and improving AI. In a recent four-day open data science conference I attended in the U.S., many sessions centered on enhancing the performance and accuracy of machine learning and large language model (LLM) algorithms. Despite the unfortunate proliferation of acronyms and technical terms in new technologies, clear distinctions and sit uating research and practice in proper contexts are crucial in the ever-evolving landscape of AI.

Interest in AI has surged with the introduction of ChatGPT, a generative AI that responds to natural language prompts. Bard from Google and LLaMa from Meta quickly followed. LLaMa, a collection of foundational language models released as open source, boasts a parameter range of 7B–65B. This implies more prediction models will proliferate without due consideration of model explainability or domain knowledge. In less than two years, OpenAI, the creator of ChatGPT, has commercialized various ChatGPT versions with new powerful features. Major tech companies, such as Google and Microsoft are consistently integrating AI into their core tools and services.

On OpenAI’s demo day in 2023, the creators of ChatGPT showcased how developers, including non-programmers, can easily build a chatbot or application using natural language text or speech commands and deploy it as a production or service tool. While OpenAI and Microsoft are dedicated to commendable missions, prioritizing safety and sustainability, it’s crucial to note that these tools are not available for free. Despite the availability of numerous open-source codes for learning and implementing AI, the landscape is predominantly shaped by major tech companies, leading innovations and changes in their pursuit of building artificial general intelligence (AGI) agents.

A positive development is the increasing discussion among governments, universities, and academic communities on collaborative efforts for sustainable development and education of AI. The Global Research on AI in Learning and Education (GRAILE) project is a good example, building a consortium of universities in the U.S. and Australia to exemplify AI-first universities sharing resources and research on AI, analytics, and digitization (for more, see https://graile.ai). In the HRD community, a growing interest in AI is apparent, highlighted in a recent podcast from the AHRD Masterclass series where three scholars discuss their research on AI (Short, 2023). In the episode, Alexandre Ardichvili distinguishes between artificial general intelligence (AGI), an aspirational concept capable of performing any task or transitioning across domains, and artificial narrow intelligence (ANI), which is designed for specific tasks or domains. Ardichvili emphasizes that ANI is already widely integrated into our daily lives.

AI, as a domain, develops smart systems and machines to perform tasks matching or exceeding human intelligence based on natural language processing, visual and spatial perception, knowledge representation, intelligent robotics, and automatic programming and reasoning. Machine learning (ML), a subfield of AI, focuses on creating algorithms that learn from data and make decisions based on patterns grounded in the data. ML often requires the final decision from humans, utilizing various linear and non-linear models for prediction, classification, recommendations, and outlier detection. Recently, the use of deep learning using artificial neural networks and brain-like logic has increased in ML (Janiesch et al., 2021). Algorithms that run these solutions are results of applying linear algebra, calculus, probability, and statistics, and surprisingly, generative AI models are no different. For a foundational understanding of the machine learning and deep learning history and concept underlying these technologies, I recommend these two easy-to-follow books (Alpadin, 2021; Kelleher, 2019).

Shifting the focus to people analytics, discussing it without its connection to AI or computational approaches can lead to a nearsighted view and ideological claims. Advancements in AI stem from academic research and visionary practitioners’ industrious application of mathematical, statistical, and ML algorithms over half a century. People analytics, although relatively new compared to the history of AI, is grounded in efforts to better measure, explain, and predict workforce and people decisions utilizing better data and research design.

My scholarly interests in technologies date back to working as an information technology (IT) project manager in the late 1990s and early 2000s. The office I worked at as a graduate assistant at the University of Illinois built a course/learning management system that ran an online master’s degree program in HRD. Technologies that comprised the backbone of the delivery platform included much simpler technologies than today, such as relational database management systems (RDBMs), server scripting languages, web and intranet servers, and a moderate list of protocols for file transferring and access management. Now, technology stacks are far more complicated in capturing, storing, extracting, and retrieving data.

Fortunately, visionary scholars in various social science disciplines foresaw the growing importance of data centric approaches and created a discipline called computational social science (CSS) about two decades ago (Lazer et al., 2020; Contractor, 2019). My scholarly journey on computational methods started with studying social network analysis (SNA) after teaching intermediate and advanced web-based learning courses on web 2.0 and dynamic web publishing for about a decade. My IT project management background was helpful in tackling self-studying machine learning, text mining, and data science approaches. Conducting several projects and self-directed learning exposed me to the importance and potentials of data analytic approaches to HRD research and practice, and I presented my synthesis of data analytical approaches and people analytics for the HRD audience to the Human Resource Development Quarterly journal (Yoon, 2018, 2021).

Upon seeing those editorials, Yonjoo Cho, the immediate past chief editor of HRDR, invited me to edit a special issue on people analytics in spring 2022. Initially hesitant due to several special issues from other fields having already covered HR analytics, workforce analytics, and human capital analytics, I reconsidered after recognizing the importance of this work. Pursuing my passion for revealing the ‘true’ values and impact of HRD work, leveraging technologies as a means to better ends in developing people, was the field I chose as a career. The call for proposals received 21 quality submissions. Some high-potential ones had to be declined due to an overlap in topics, and discussions of AI were not directly tied to people analytics. This year, I also observed many poster sessions and research abstracts for AI-related topics at the 2023 AHRD conference in Minneapolis, USA. I mention these because I understand that a handful of scholars have been conducting research on technologies, including AI and people analytics in HRD. PA can undoubtedly incorporate various technological and data analytical approaches including AI I mentioned earlier.

Importantly, PA is not merely about data analysis; nor should it solely revolve around the application of AI or computational methods. As Ardichvili pointed out, ANIs are already prevalent, and PA is widely practiced for attracting, recruiting, developing, and maintaining talent. It is rapidly advancing and evolving with the progress in AI. However, data analytical approaches, such as PA must explicitly leverage the rich tradition of qualitative, quantitative, and mixed methods that have amassed robust research methodologies and ethical principles. Many presenters at the data science conference I attended have emphasized the need for more qualitative research associated with the deployment of computing algorithms.

Adding the term “people” in front of analytics signifies the critical importance of applying analytics with a focus on developing and empowering people. While algorithmic decision-making, real-time processing, unstructured, and large data might be inevitable trends, workforce and people data remain comparatively small. The design and analysis of research projects examining stories, experiences, structured and tabular data, longitudinal data, and explainable and responsible AI will continue to be critical, necessitating the development and use of domain-based logic models (Levenson & Fink, 2017).

With these considerations, the five articles included in this special issue are the result of year-long efforts from scholars and practitioners of diverse backgrounds. The special issue’s call for papers invited scholars from HRD and related fields (HR, management, and organization) to contribute manuscripts focusing on conceptual and theoretical frameworks and methodological advancements related to people analytics. The following questions are what I envision readers of this special issue would ask, and the five articles included provide excellent guidance:

• What is people analytics? Is it a new research method or a growing area of new organizational practices? What roles or opportunities does PA provide for HRD research and practice?

• PA research has been growing and driven by a management and performance lens, lacking human and organizational development perspectives. What roles does HRD have?

• As new technologies, such as cloud computing, unstructured data, and AI, are increasingly used by workers and organizations, how should HRD research and practice leverage them, and with whom should HRD collaborate more?

• How can people analytics be utilized by HR and organization scholars to better understand, and leverage established and emerging research methods?

The first paper, by Rasmussen, Ulrich, and Ulrich, is entitled “Moving People Analytics from Insights to Impacts.” This article is a sequel to Rasmussen and Ulrich’s (2015) influential work, “Learning from Practice: How HR Analytics Avoids Being a Management Fad.” Drs. David Ulrich and Thomas Davenport are recognized for helping elevate the status of data analytics in the HR and management space respectively. Davenport and colleagues’ work was influential in prompting organizational leaders and data scientists to recognize the value of applying data analytics to talent development (Davenport et al., 2010). In 2015, Rasmussen and Ulrich emphasized how HR analytics often take an inward-looking approach, focusing on HR issues. To be impactful, they emphasized, it must take an outside-in approach, focusing on solving critical business problems leveraging technologies. This proposition was adopted as the most important factor for PA success in the practice community. In this updated article, Rasmussen and Ulrich present a framework for designing and implementing PA in an organization based on creating specific values for key stakeholders. Based on successful multiple use cases, their suggestions for targeting multi-level outcomes for both individuals and teams, balancing between moonshot and tangible projects, and integrating both HRM and HRD solutions provide a useful framework for PA processes.

The second and third papers are literature review studies that used different methods. Reading them together is strongly recommended. The second article by Yoon and colleagues applied a bibliometric review to 159 articles combining topic modeling and clustering analysis. Their approach to selecting literature is noteworthy. To review the PA literature from the HRD lens, they combined multiple analytics terms with HRD topics and interventions where assessments and measurements are frequent or critical. They also added a manual review of HRD publications on data analytics and AI. From topic modeling, they captured multiple topics of PA research around five emerged themes: (1) workforce/HR management and planning, (2) ethics, fairness, bias, and societal impacts, (3) data, methods, capability, and technology adoption, (4) analytics for HR functions, and (5) analytics in healthcare and start-up companies. Clustering analyses added that the majority articles talk about HR and organizational transformation and ethics, and empirical articles reporting impacts on tasks and HR functions were much less indicating strong needs for more empirical evidence. They also discuss which topics in the current PA research landscape are the primary domain of HRD research, and how further integration and collaboration between HRD and HRM is in great order.

The third paper by Lee and Lee has applied an integrative literature review synthesizing 91 articles retrieved from searching PA related terms. Their first part captured the current knowledge of PA research based on (a) comparing terms and definitions, (b) presenting five core building blocks of strategic alignment, analytic competency, data, technologies, and organizational support, (c) types of data analytics (descriptive, predictive, and prescriptive), and (d) major ethical issues and considerations. The second part of the article discusses how PA can be applied to HRD research and practice in four core domains of individual development, career development, organization development, and organization performance based on Gilley et al.’s (2002) view of HRD work contexts. To do so, they emphasize HRD research to build upon established behavioral and measurement frameworks and collaboration between practitioners and academic researchers.

The fourth paper by Chang and Ke tackles the important issue of ethics and responsible use in AI. They have conducted an integrative literature review of 75 articles and incorporated three established sustainability frameworks: corporate social responsibility (CSR), environment, social, and governance (ESG), and the United Nations’ sustainable development goals (SDGs) into a framework called SRAI (Socially Responsible Artificial Intelligence). Their proposed framework has captured the essential building blocks and their alignment through five distinct hierarchical levels: economic (robust AI), legal (lawful AI), ethical, philanthropic (human-centered AI), and environmental (sustainable AI). Each level also highlights how project, people, and planet-centered use of AI can effectively address SRAI principles. The preceding articles highlight the multi-dimensional and critical nature of ethics in PA and highlight the relevance of HRD work, and the proposed SRAI framework can be a useful guide for studying and better assessing AI sustainability.

In the fifth paper, Yoo and colleagues demonstrate the application of network analysis, a major method in computational social science, to test and advance the theory of social learning, team learning, and organizational learning. The 3 E framework captures with whom individuals explore for new knowledge and resources outside the team (Exploration), seek information and resources (Exploitation), and restructure or repurpose them (Exaptation). Social network analysis (SNA) enables researchers to examine structural patterns and influences based on horizontal or vertical nodes (actors and/or resources) interdependence (Borgatti et al., 2022). In SNA, levels such as individuals, teams, and the organization can be captured as actor or node attributes, while interactions and connectivity among actors are captured by multiplex relationships (e.g., ties of trust, information seeking, friendship, knowledge sharing, etc.). Using a case scenario and code examples, Yoo and colleagues illustrate how the phenomena of workplace learning can be empirically examined, applying SNA. As a result, one can assess the cohesion of various networks and identify central players in a multiplex and multi-level learning network. This study showcases an exemplary application of computational methods for theory testing and advancements.

While the special issue covers valuable ground, acknowledging certain limitations such as space constraints and methodological choices, the included articles present multiple conceptual and implementation frameworks. These encompass conducting PA projects for outcomes and impacts (Rasmussen et al., forthcoming), addressing ethical and responsible use of AI (Chang and Ke, forthcoming), applying network analysis to workplace learning (Yoo et al., forthcoming), applying PA methods and topics to HRD core domains (Lee and Lee, forthcoming), and discussing the current research landscape and future research needs (Yoon et al., forthcoming). The recent editorial from the chief editor, Thomas Garavan, underscores how HRDR has advanced the field of Human Resource Development (HRD) through the development of new theories, methodologies, and philosophical perspectives (Garavan, 2023). Armed with concepts and methods that emphasize people, technologies, data, and business priorities, PA offers HR and organization scholars with innovative and data-grounded approaches to align these essential elements for strategic and evidence-based impactful HR.