Firms are integrating AI across core functions—HR, marketing, logistics and finance—to automate routine work, sharpen analytics and personalize interactions, but governance gaps and weak organizational capabilities risk uneven, irresponsible implementation.

Main Finding

A PRISMA-style systematic review (Web of Science + Scopus, through Dec 2025) of 160 peer‑reviewed articles finds common patterns of AI adoption across core firm functions (human resources, marketing/customer service, logistics/manufacturing, finance). AI chiefly (1) enhances analytics and decision support, (2) automates routine tasks, (3) personalizes customer interactions, and (4) raises cross‑functional governance, capability and socio‑technical alignment challenges. The paper proposes an integrative conceptual model and research propositions for function‑level implementation and responsible deployment.

Key Points

• Scope and orientation

  • Systematic literature review of AI implementation within internal organizational functions (160 articles through Dec 2025).
  • Emphasis on functional deployment (vs. sectoral or technology‑only reviews); acknowledges surge in chatbots/LLMs since 2021.

• Cross‑cutting theoretical anchors

  • Dynamic Capabilities (sensing, seizing, reconfiguring resources).
  • Chandlerian functional coordination and multilevel socio‑technical alignment.
  • Technology Acceptance Model (employee acceptance at individual level).

• Human Resources

  • Major applications: scheduling/timetabling, workforce planning, recruitment screening, automated assessment of training needs, workplace climate monitoring (harassment, bullying detection), and adaptive training (virtual tutors, educational chatbots).
  • Benefits: efficiency, reduced admin burden, better allocation of human time to higher‑value tasks.
  • Risks/needs: algorithmic bias, transparency, oversight, training deficits for employees.

• Marketing & Customer Service

  • Major applications: active listening/real‑time customer analytics, predictive personalization (recommendation engines), chatbots/virtual assistants for interaction and lifecycle management.
  • Benefits: more dynamic customer insight, personalized offers, faster service, brand enhancement.
  • Trends: advances in NLP and voice interaction; customer service data feeding cross‑functional processes.

• Logistics & Manufacturing

  • Major applications: Industry 4.0 integration (sensors, IoT), robot control, automated visual inspection, predictive maintenance, demand forecasting, inventory/transport optimization, supply‑chain risk modeling (including generative AI scenario analysis).
  • Benefits: process reliability, cost reductions, improved resilience to disruptions.
  • Needs: alignment between operational needs and available technologies; integration across cyber‑physical systems.

• Finance & Operations

  • Major applications: large‑scale data analysis for fraud/risk assessment, process automation, voice recognition in workflows, decision support for financial planning.
  • Benefits: enhanced risk modelling, process efficiency.
  • Cautions: data quality, model governance, regulatory compliance.

• Cross‑functional challenges highlighted

  • Governance and responsible AI (bias, transparency, regulation).
  • Organizational capabilities (data governance, analytics skills, reconfiguration capacity).
  • Socio‑technical alignment (matching artefacts, structures, and human roles).
  • Need for validated models and more empirical robustness (many studies remain exploratory).

Data & Methods

• Literature search: Web of Science and Scopus databases, coverage up to December 2025.
• Methodology: Systematic review following established guidelines (PRISMA reporting is referenced).
• Sample: 160 peer‑reviewed articles met inclusion criteria (focus on AI implementation in organizational functions).
• Outputs: synthesis of functional‑level patterns, development of an integrative conceptual model and research propositions.
• Limitations noted by authors: preponderance of exploratory studies; fewer validated empirical models; rapid post‑2021 developments (LLMs/chatbots) create a moving target.

Implications for AI Economics

• Labor and task allocation

  • AI reshapes task composition across functions: routine task automation frees labor for higher‑order work but may generate heterogenous displacement and creation across roles (skill‑biased reallocation).
  • Research directions: quantify function‑specific employment and wage effects; measure task complementarities between AI and human labor.

• Productivity and complementarities

  • Potential productivity gains via improved forecasting, personalized marketing, and automation.
  • Research directions: estimate causal effects of function‑level AI adoption on firm productivity, margins, and innovation using panel data, diff‑in‑diff, IVs, or field experiments.

• Organizational boundaries and transaction costs

  • AI‑enabled information flows may change internal coordination costs and the optimal allocation of activities across departments or firms (vertical/horizontal boundaries).
  • Research directions: evaluate how AI affects outsourcing vs. insourcing decisions and inter‑departmental spillovers.

• Investment, skills, and distributional effects

  • Adoption requires investments in data infrastructure, governance, and human capital (training, reskilling).
  • Research directions: value of training investments; returns to data infrastructure; heterogeneity by firm size and industry.

• Risk, governance, and regulation

  • Algorithmic bias, opacity, and model risk create externalities and potential regulatory interventions that affect adoption incentives.
  • Research directions: model regulatory impacts on adoption; quantify costs of poor governance (litigation, reputation); design policy to align incentives for responsible AI.

• Supply‑chain resilience and macro‑externalities

  • AI can improve forecasting and resilience, with economy‑wide implications (reduced volatility, altered demand shocks propagation).
  • Research directions: macro/sectoral assessment of AI‑driven resilience; welfare consequences.

• Measurement and empirical strategy suggestions

  • Use firm‑level administrative data augmented with function‑level adoption indicators (software purchases, procurement, job descriptions).
  • Employ quasi‑experimental designs (staggered rollouts, policy changes) and process‑level metrics (throughput, error rates) to identify effects.
  • Explore matched comparisons, instrumental variables capturing differential exposure to relevant data/skills, and randomized pilots for causal inference.

Overall, the paper documents convergent functional patterns of AI use and frames open empirical questions central to the economics of AI adoption: how gains are distributed across functions and workers, what complementarities drive productivity, how governance shapes outcomes, and how to measure causal impacts.