Generative AI is reshaping organizational decision-making into six distinct roles but rarely makes the final call; firms use GenAI extensively for attention, intelligence, design and implementation tasks, yet integration into the choice stage remains limited.

Rethinking organizational decision-making: The emerging roles and tasks of generative artificial intelligence

Niklas Schulte, Dominik K. Kanbach · Fetched July 13, 2026 · Management Review Quarterly

semantic_scholar review_meta n/a evidence 7/10 relevance Full text usable extracted full text DOI Source PDF

A systematic review of 68 studies identifies 53 GenAI tasks (grouped into 18 categories), proposes six active GenAI roles plus one collaborative human–AI role, and finds that GenAI is widely used across decision-making stages but is fragmented and underutilized in the choice phase.

The rapid diffusion of generative artificial intelligence (GenAI) has triggered a transformative shift in how organizations approach decision-making. Despite growing enthusiasm and widespread adoption across industries, GenAI’s specific tasks and roles, and the ways in which they shape the interplay of human cognition and algorithmic enhancement in organizational decision-making, remain insufficiently understood. Addressing this gap, this study conducts a systematic literature review that identifies 68 relevant publications to synthesize and advance current knowledge on the integration of GenAI into decision-making. The study identifies 53 tasks performed by generative applications, aggregates them into 18 task categories, and maps these tasks and categories onto six recursive decision-making components: attention, intelligence, design, choice, implementation, and feedback. Building on the harmonization and translation of these tasks, we propose a typology comprising six active GenAI roles and one collaborative human-AI role. We then develop a processual framework that specifies how and when GenAI is embedded within organizational decision-making processes, delineating how generative applications support, augment, or co-perform decision-making activities. Our findings reveal a fragmented application landscape and highlight the limited integration of GenAI in the choice phase of organizational decision-making. By offering a structured typology and a processual conceptual framework, this study clarifies the evolving interplay between human decision-makers and generative technologies. In doing so, it provides a foundation for theory-advancing research and for more explicit and actionable managerial practice.

Summary

Main Finding

A systematic literature review of 68 publications shows that generative AI (GenAI) is already performing a wide range of decision-related tasks in organizations (53 tasks grouped into 18 categories) and can be mapped onto six recursive components of organizational decision-making (attention, intelligence, design, choice, implementation, feedback). The authors propose a typology of six active GenAI roles plus one collaborative human–AI role and a processual framework describing how GenAI supports, augments, or co-performs decision activities. Key empirical/theoretical takeaways: GenAI applications are fragmented across decision stages, tend to alleviate bounded rationality selectively (especially improving formal/procedural aspects), and are under-integrated in the choice phase; governance, verification, data assets, and human oversight remain critical.

Key Points

Scope and synthesis
- Systematic literature review (Tranfield et al. approach) of 68 relevant articles synthesizing GenAI tasks/roles in organizational decision-making.
- Identified 53 discrete GenAI tasks, aggregated into 18 task categories, and mapped across six decision-making components.
Decision-making components used for mapping
- Attention, intelligence, design, choice, implementation, feedback (building on Simon; Eisenhardt & Zbaracki; Ocasio; Greve).
Typology and roles
- Six active GenAI roles + one collaborative human–AI role (authors propose role definitions and link them to task clusters and decision stages).
Empirical patterns and limitations
- Fragmented application landscape: GenAI is widely used for tasks like information summarization, scenario generation, forecasting, and solution ideation, but less so for final choice/selection.
- Selective alleviation of bounded rationality: GenAI often enhances formal/ procedural rationality (efficiency, information processing, search), less clearly improving substantive/ normative choice quality.
- Cognitive and socio-material effects: GenAI shifts attention allocation, alters cognitive effort and creative diversity, and creates hybrid human–AI cognition and coordination challenges.
- Risks and governance needs: stochastic outputs, hallucinations, overconfident rationales, and prompt/user biases create verification, provenance, and accountability requirements.
Theoretical contributions
- Advances understanding of algorithmic agency, socio-material entanglement, and hybrid human–AI cognition within organizational decision theory.
- Calls for revisiting attention-based views and bounded rationality in light of generative decision support.

Data & Methods

Method: Systematic literature review following a replicable sequence (Tranfield et al. 2003).
Corpus: 68 publications deemed relevant to GenAI in organizational decision-making (selection and synthesis procedures reported in the paper).
Analytical outputs:
- Catalogued 53 tasks performed by GenAI and grouped them into 18 task categories.
- Mapped tasks and categories onto six decision-making components: attention, intelligence, design, choice, implementation, feedback.
- Developed a six-role (plus one collaborative) typology for GenAI in decision-making and a processual framework showing when/how GenAI supports, augments, or co-performs activities.
The review integrates prior decision-making and DSS literatures (Simon; Eisenhardt & Zbaracki; Ocasio; literature on DSS, predictive AI, and socio-material views).

Implications for AI Economics

Firm-level productivity and value creation
- GenAI can raise effective decision-processing capacity (search, summarization, scenario generation), implying potential productivity gains. But gains are stage-specific: larger for intelligence/design/implementation tasks, smaller or riskier for final choice without governance.
- Returns to GenAI investments are likely conditional on complementary assets: data quality, prompting practices, verification processes, and managerial governance.
Labor, skills, and the division of cognitive labor
- Expect shifts in labor demand toward skills that complement GenAI (prompt engineering, oversight, verification, governance, strategic judgment) and away from routine information-processing tasks.
- Hybrid cognition implies reallocation rather than pure substitution for many decision tasks; economic models should treat GenAI as a partial substitute/complement depending on task and decision stage.
Market structure and competitive advantage
- Heterogeneous adoption and differences in complementary assets (data, governance, domain expertise) can amplify firm heterogeneity and first-mover advantages, potentially increasing concentration in some sectors.
- Fragmented application landscape suggests uneven realized gains across firms and industries—empirical work should measure dispersion.
Measurement and welfare
- Welfare/productivity measurement must account for (a) stage-specific effects (attention/intelligence vs. choice), (b) costs of verification and governance to mitigate hallucinations and overconfidence, and (c) potential negative externalities from erroneous or biased outputs.
- Benefits from GenAI may be overstated if studies do not account for the transaction and monitoring costs needed to make generative outputs decision-grade.
Policy and regulation
- Because GenAI can materially influence organizational choices and attention allocation, regulators and policymakers should be attentive to accountability, provenance, and auditability requirements—especially in high-stakes decision domains.
Research directions for AI economics
- Structural and causal estimation: quantify productivity returns to GenAI by decision stage and identify complementarities with data/governance investments.
- Models of task allocation: formalize how firms optimally allocate decision tasks between humans and GenAI across attention, intelligence, design, choice, implementation, feedback.
- Adoption and diffusion: study heterogeneity in adoption, complementarities, and implications for market structure and wage distribution.
- Measurement of costs: estimate verification/monitoring costs and the economic impact of hallucination/uncertainty in generative outputs.
- Welfare and regulation: evaluate social welfare implications and design policies (standards, liability rules, transparency mandates) that balance innovation and risk mitigation.

Limitations noted by the authors (relevant for empirical work) - The study is a literature synthesis (no new empirical dataset); findings reflect the state of published research (corpus up to the review period). - Evidence is fragmented and often qualitative or experimental; economics-oriented empirical validation (firm-level, panel, RCTs, structural models) is still needed.

Assessment

Paper Typereview_meta Evidence Strengthn/a — This is a systematic literature review and conceptual synthesis rather than a primary empirical study that estimates causal effects; it aggregates findings from diverse sources but does not provide new causal identification. Methods Rigormedium — The study uses a systematic search and synthesizes 68 publications, maps tasks into categories, and proposes a typology and processual framework—showing careful qualitative coding and conceptual integration; however, rigor is limited by potential selection and publication bias, subjective coding decisions, heterogeneity of included studies (conceptual papers, case studies, limited causal empirical work), and the quickly changing GenAI literature. SampleA systematic sample of 68 publications on generative AI in organizational decision-making (likely a mix of peer‑reviewed articles, conference papers, working papers, and practitioner pieces) covering multiple industries and methods (conceptual, qualitative case studies, and some empirical work); exact date range and inclusion criteria not specified here. Themeshuman_ai_collab org_design adoption productivity GeneralizabilityLimited to published literature and the authors' inclusion criteria (publication and selection bias)., Heterogeneous study types and contexts reduce ability to generalize findings across firms, industries, or countries., Rapid evolution of GenAI tools may outdate mapped tasks and roles quickly., Predominance of conceptual and qualitative studies limits inference about frequency or economic impact across the broader economy.

Claims (10)

Claim	Direction	Outcome	Confidence & Evidence	Details
This study conducts a systematic literature review that identifies 68 relevant publications. Research Productivity	positive	number of publications included in the literature review	Reading fidelity high Study strength high	n=68 0.4
The study identifies 53 tasks performed by generative applications. Task Allocation	positive	count of tasks performed by GenAI	Reading fidelity high Study strength high	n=68 53 tasks 0.4
Those 53 tasks are aggregated into 18 task categories. Task Allocation	positive	number of task categories	Reading fidelity high Study strength high	n=68 18 categories 0.4
The study maps these tasks and categories onto six recursive decision-making components: attention, intelligence, design, choice, implementation, and feedback. Organizational Efficiency	positive	alignment of tasks/categories to decision-making components	Reading fidelity high Study strength medium	n=68 0.24
Building on the harmonization and translation of these tasks, we propose a typology comprising six active GenAI roles and one collaborative human-AI role. Task Allocation	positive	number and type of GenAI roles proposed	Reading fidelity high Study strength medium	n=68 six active roles and one collaborative human-AI role 0.24
We develop a processual framework that specifies how and when GenAI is embedded within organizational decision-making processes, delineating how generative applications support, augment, or co-perform decision-making activities. Organizational Efficiency	positive	modes and timing of GenAI embedding in decision processes	Reading fidelity high Study strength medium	n=68 0.24
Our findings reveal a fragmented application landscape for GenAI in organizational decision-making. Adoption Rate	negative	degree of fragmentation/heterogeneity in application of GenAI	Reading fidelity high Study strength medium	n=68 0.24
The study highlights the limited integration of GenAI in the choice phase of organizational decision-making. Task Allocation	negative	extent of GenAI integration in the choice phase	Reading fidelity high Study strength medium	n=68 0.24
GenAI’s rapid diffusion has triggered a transformative shift in how organizations approach decision-making. Organizational Efficiency	positive	impact of GenAI diffusion on organizational decision-making practices	Reading fidelity medium Study strength low	not reported 0.07
By offering a structured typology and a processual conceptual framework, this study provides a foundation for theory-advancing research and for more explicit and actionable managerial practice. Research Productivity	positive	utility of the study's outputs for future research and managerial practice	Reading fidelity high Study strength speculative	n=68 0.04