Mapping the evolution of algorithmic HRM (AHRM): a multidisciplinary synthesis

The fourth industrial revolution is driving pervasive digital transformation across industries, accelerating changes in business processes and systems and increasingly impacting HRM strategy and practices. Organizations are adopting algorithmic tools in HRM, ushering in an era of Algorithmic HRM (AHRM) with significant implications for productivity, role redesign, upskilling, and change management. To respond to these trends, the authors conduct a multidisciplinary synthesis of algorithm-related concepts to produce a General Framework for Algorithmic Decision-Making, which then anchors a focused mapping of AHRM’s evolution and a Framework for Algorithmic AHRM Tools. This offers conceptual clarity and distinguishes automated from augmented HR decision-making and their implications for strategic HRM (SHRM). Prior research links digital transformation, strategy, and SHRM through the resource-based view, emphasizing human capital and technologies as strategic assets. While SHRM research often takes macro perspectives on human capital investments to improve firm performance, effective deployment of AHRM requires sophisticated technologies. From a human–AI interaction perspective, the literature notes “low-status” automation and “high-status” augmentation and stresses the coexistence of humans and AI in workplaces, including HR ecosystems that influence employee experience and engagement. However, the broader construct of “digital HRM” has lacked clarity, conflating HR analytics, algorithms, platforms, AI, and big data under one umbrella. Adopting the term AHRM—“the use of software algorithms that operate on the basis of digital data to augment HR-related decisions and/or to automate HRM activities”—the field emphasizes big data generation and use, software algorithmic processing, and partial or full automation of HR decisions. This study’s dual objectives are: (1) advance conceptual clarity for AHRM, and (2) map AHRM’s evolution from 2000 to 2022. The paper reviews prior algorithmic HRM reviews, develops a theoretical framework synthesizing AHRM definitions with an AI taxonomy, then applies a science mapping methodology to chart the literature and present findings, implications, limitations, and future research directions.

A retrospective review of technology and HRM (1961–2019) identified thematic continuity in how technology affects jobs, supports HR practices and decisions, and manages technology workers, with landmark publication peaks aligning with personal computers (1977) and consumer internet (1997). Recent reviews (2019–2023) collectively evidence growing interest in AI and related technologies in HRM but vary widely in scope, databases, and methods. Four reviews survey AI in HRM broadly across recruitment, performance, remuneration, training and development, HR analytics, and e-HRM. Two focus on blockchain: one on systems-level HR implications (e.g., authentication in recruitment) and one on detecting resume/candidate fraud. Others focus on tactical HRIS applications; machine learning objectives and their HR uses; big data-driven HR practices; intelligent automation technologies (AI, robotics, smart tech) and HRM; and AI within sustainable development business models (with limited HR findings). Together these studies illustrate a multidisciplinary, fragmented landscape with diverse time ranges and foci, underscoring the need for a unifying map of AHRM. Based on this, the authors pose RQ1: What are the dominant, emerging, and state-of-the-art themes in published research from 2000 to 2022 that examine organizational tools for AHRM?

The study employs a theory-led science mapping (scientometric) approach, guided by the General Framework for Algorithmic Decision-Making (synthesizing Meijerink et al.’s AHRM definition with Herrmann’s AI taxonomy). Database selection favored Scopus due to broad coverage and export capabilities; Google Scholar was excluded due to grey literature and lack of bulk export, and Web of Science has narrower journal coverage for AI relative to Scopus. Search strategy (Scopus syntax, run 17 May 2023): ((AUTHKEY ("human resource" OR "personnel") AND TITLE-ABS-KEY-AUTH ("decision support" OR "expert system" OR "robotic process automation" OR "data science" OR "business intelligence" OR "machine learning" OR "deep learning" OR "big data" OR algorithm* OR "algorithmic manage*" OR metaheuristic* OR autonom*))) AND (PUBYEAR > 1999) AND (LIMIT-TO (SRCTYPE, "p") OR LIMIT-TO (SRCTYPE, "j") OR LIMIT-TO (SRCTYPE, "k") OR LIMIT-TO (SRCTYPE, "b")) AND (LIMIT-TO (DOCTYPE, "cp") OR LIMIT-TO (DOCTYPE, "ar") OR LIMIT-TO (DOCTYPE, "ch") OR LIMIT-TO (DOCTYPE, "bk")) AND (EXCLUDE (PUBYEAR, 2023)) AND (LIMIT-TO (LANGUAGE, "English")). Inclusion criteria restricted to peer-reviewed journals, books/chapters, conference papers/proceedings; years 2000–2022 (reflecting mainstreaming of algorithmic AI with hardware advances, optimization algorithms, open-source libraries, public datasets). Initial retrieval yielded 1675 publications; verification of titles/abstracts/keywords for the top 100 cited reduced false positives; deduplication removed 3 entries (Scopus indexing issues), resulting in 1672 publications: journals 58%, conference papers/proceedings 38%, books/chapters 4%. Science mapping used SciMAT v1.1.05 for preprocessing and evolutionary mapping across four periods (2000–2012; 2013–2017; 2018–2020; 2021–2022). Keyword co-occurrence analysis applied a minimum threshold of two occurrences across title, abstract, and keywords to reduce noise. Clustering produced thematic networks per period; cross-period linkages (solid lines = shared ≥2 keywords; dashed = 1 keyword) and their thickness indicate relationship strength; cluster sphere sizes reflect h-index. The method aligns with prior AHRM bibliometric studies but adds a longitudinal evolutionary map of themes.

• Corpus size and growth: 1672 AHRM publications (2000–2022). Publications were under 20/year until 2006, then steadily rose through 2019; output doubled in 2020 vs. 2018/2019, driven by HR scheduling, allocation, and recruitment applications. Distribution: 58% journal articles, 38% conference papers/proceedings, 4% books/chapters. - Multidisciplinarity: Subject area distribution indicates AHRM is predominantly published outside HRM/business: Computer Science 24%, Engineering 15%, Business/Management/Accounting 11%, with additional contributions from Decision Sciences (8%), Mathematics (8%), Medicine (7%), Physics and Astronomy (3%), Economics/Econometrics/Finance (2%), and others (14%). - Evolutionary themes (Fig. 5): Three thematic categories emerged: (1) AHRM technologies (data science, algorithmic AI, metaheuristics, tracking); (2) Specialized applications (learning and development; workforce planning; allocation and scheduling; recruitment); (3) Multiple and integrated contexts (decision theory; augmented decisions; Industry 4.0 automation). Early influence (2000–2012) from decision theory, learning and development, and data science linked over time to workforce planning, allocation, and scheduling via algorithmic AI and metaheuristics. Amber area showed employee tracking linked to Industry 4.0 and recruitment automation. - Cluster details and influence: DECISION-THEORY (2000–2012) displayed dense interdisciplinary technical themes (e.g., artificial intelligence, rough sets, fuzzy logic) with recruitment and allocation/scheduling often integrated; h-index 25. AUGMENTED-DECISIONS (2013–2017) added performance and risk management with strong metaheuristic influence; HRIS context; h-index 22. INDUSTRY-4.0-AUTOMATION (2018–2020) was less developed (lower density) with performance management, personnel security, safety, embedded systems; h-index 10. - Terminological confusion: Persistent conflation of AI, ML, big data, and HR umbrella terms (AHRM, digital HRM, HR analytics, e-HRM) indicates lack of precision and construct clarity; a broad, integrated AHRM definition can help align technologies with HR activities and SHRM accountability. - Publication venues: Only 11% of AHRM publications appear in business/HRM journals, suggesting limited uptake by HRM scholars and underscoring the need for multidisciplinary collaboration between HR and technical fields. - Emphasis on non-strategic functions: Most studies focus on operational HR functions (scheduling/rostering, recruitment/selection, learning and development), with fewer integrative SHRM applications. Exceptions include workforce planning and HR analytics, and integrated contexts (recruitment, allocation and scheduling). Many studies adopt a tools-view with single-objective applications; fewer multi-objective studies remain largely non-strategic. - Automation vs augmentation in specialized applications (Table 2 synthesis): Recruitment is the most automated/prescriptive AHRM tool, involving both automated and augmented modes; metaheuristics are emerging in recruitment. Allocation & scheduling and workforce planning show both augmented decisions and Industry 4.0 automation, supported by expert/decision support systems and algorithmic AI. Learning & development is primarily augmented, supported by expert systems and algorithmic AI. - Framework contribution: Integration of data-driven mapping (Table 2) with the theory-driven analytics hierarchy yields a General Framework for Algorithmic HRM Tools (Fig. 9) distinguishing augmented (descriptive/diagnostic/predictive analytics with humans in the loop) from automated (prescriptive analytics, RPA, autonomous systems) HR decision-making.

The study addresses RQ1 by identifying dominant, emerging, and state-of-the-art themes in AHRM from 2000–2022. Dominant themes include algorithmic AI and data science underpinning specialized applications in allocation and scheduling, recruitment, learning and development, and workforce planning. Emerging themes include metaheuristics in recruitment and broader Industry 4.0 automation influences, as well as integrated contexts linking recruitment with allocation/scheduling and, later, performance and risk management. The findings highlight that while algorithmic tools have been widely applied to operational HR tasks, there is a relative paucity of integrative SHRM-focused applications, suggesting an opportunity to extend AHRM beyond tactical efficiencies to strategic human capital decision-making. Conceptual confusion around AI-related terminology in HRM underscores the need for a unified AHRM framework to align technologies with HR decisions and clarify automated versus augmented modes. The multidisciplinary publication pattern indicates that much AHRM development is driven by computer science and engineering, implying HRM researchers and practitioners would benefit from closer collaboration with technical experts to design ethically robust, explainable, and accountable systems. Practically, HR practitioners should form cross-functional project teams with IT and data analysts to select and implement AHRM tools that support both operational efficiencies and strategic objectives. The frameworks presented guide how to position technologies within analytics hierarchies and decision modes, offering pathways to integrate predictive and prescriptive tools into SHRM while maintaining human oversight where appropriate.

This study provides conceptual clarity for AHRM by synthesizing a General Framework for Algorithmic Decision-Making and, building on a large-scale science mapping of 1672 publications (2000–2022), proposing a General Framework for Algorithmic HRM Tools that distinguishes augmented from automated HR decision-making. The review reveals AHRM’s multidisciplinary nature, with substantial contributions from computer science and engineering, and clusters the literature into AHRM technologies, specialized HRM applications, and multiple/integrated HRM contexts. Current research emphasizes specialized applications (recruitment, workforce planning, allocation and scheduling, learning and development) with limited focus on integrative SHRM contexts. The contributions offer a roadmap for aligning analytics hierarchies and AI paradigms with HR decision modes, aiding researchers and practitioners in framing and deploying AHRM responsibly. Future research should deepen integrative SHRM applications of AHRM, develop hybrid AI approaches that enhance fairness, explainability, and accountability, and explore the expanding role of generative AI in recruitment and learning/development within rigorous ethical and governance frameworks. Multidisciplinary teams spanning HR, computer science, data science, and engineering are essential to advance theory and practice, and educational and professional development programs should build AHRM competencies for HR professionals.

• Database and coverage: Although Google Scholar is the largest bibliometric database, Scopus was used due to export constraints and grey literature noise in Google Scholar; differences in coverage may affect completeness. - Review-based evidence: The study is a secondary analysis; no primary empirical data were collected, which may limit contextual depth. - Temporal scope: Mapping covers 2000–2022; rapid innovation in 2023 onward, notably generative AI (e.g., ChatGPT), may shift the AHRM landscape. The paper notes strong potential impacts in recruitment and learning/development, alongside ethical risks (bias, fairness, privacy, security, transparency, explicability, accountability) that require focused future research. The authors propose generative AI’s implications and ethics as priority areas for AHRM research.