How Can Generative AI Impact Firm Productivity?

• Task productivity: Generative AI reduces the marginal cost of producing drafts and ideas, which increases output per unit of time and especially benefits lower skilled workers.
• Decision quality: Generative AI expands access to codified knowledge and generates higher quality informational inputs, which improves judgments and resource allocation.
• Organizational learning: Generative AI lowers the cost of codifying and sharing knowledge, which improves replication of practices and reduces coordination costs.
• Firm boundary efficiency: Generative AI changes the relative cost of internal versus external production of knowledge intensive tasks, which enables reallocation and automation.
• Human-machine complementarity: Generative AI raises productivity when workflows are redesigned to combine human judgment with machine generation.
• Coordination efficiency: Generative AI standardizes communication and maintains shared context in distributed teams, which reduces delays and errors.
• Governance effectiveness: Generative AI increases productivity only when supported by quality control, oversight routines, and clear decision rights for AI use.

Generative AI modifies how knowledge is produced, interpreted, and used in a firm. Productivity effects arise when these modifications change the relationships between information, coordination, and decision quality. What follows outlines the main mechanisms proposed in academic research, expressed in terms of their core variables and their interactions.

Generative AI alters task productivity by reducing the cost of producing drafts, summaries, and analytical elements. The model supplies initial versions of work outputs, which reduces variance in worker performance. Workers reallocate time from production to evaluation, error detection, and integration. Productivity increases through the interaction of two variables. Lower cost of initial content generation increases available time for judgment intensive work. As the complementarity between these activities strengthens, the productivity of knowledge work increases.

Generative AI influences decision quality through its effect on information availability. Decisions depend on the structure and completeness of inputs. Models reduce the cost of retrieval and synthesis of codified knowledge. Increased availability of structured information reduces forecast error and increases quality of scenario analysis. The causal link runs from higher quality inputs to improved matching of resources to objectives, which in turn raises the productivity of decisions related to planning and operations.

Generative AI shapes organizational learning by increasing the rate of knowledge codification. Codification transforms tacit experience into reusable artifacts. By reducing the cost of producing procedures, rationales, and explanations, the model increases the volume and timeliness of shared knowledge. This improves the stability of knowledge repositories, reduces dependence on individuals, and lowers coordination costs across units. As a result, variance in performance falls and replication of high quality practices increases.

Generative AI modifies the task allocation structure of the firm. When the cost of producing intermediate knowledge goods declines, the firm redesigns which tasks are performed by humans and which by automated systems. The ratio of judgment intensive tasks to routine analytical tasks shifts. Human effort concentrates on evaluation and oversight while the model performs generative steps. This reallocation increases productivity when the firm aligns capabilities with the tasks to which they add the most value.

Generative AI reinforces human machine complementarity. Research on technical change shows that technologies produce gains when they complement scarce human skills. Models complement expertise in problem framing, contextual reasoning, and strategic interpretation. Productivity gains materialize when workflows isolate tasks requiring human interpretation from tasks the model can generate. Without this redesign, potential complementarities remain unused.

Generative AI improves coordination quality by reducing ambiguity in communication. Models standardize language, provide clarifications, and maintain shared context for distributed teams. Clearer communication reduces delay and rework, and stabilizes shared mental models.

The magnitude of productivity gains depends on governance quality. Firms must implement verification routines, risk controls, and decision rights for AI use. Governance maintains reliability of outputs and prevents productivity losses due to errors in model generated content.

References

• Brynjolfsson E et al 2023 Generative AI at Work
• Noy S Zhang L 2023 Experimental Evidence on Large Language Models in Writing Tasks
• Grant R 1996 Toward a Knowledge Based Theory of the Firm
• Simon H 1997 Administrative Behavior