Federal Reserve — Working Paper 2025-053 on AI and Work

TL;DR

• Formal research into AI and labor/productivity dynamics.
• Useful for strategy and policy context around adoption impacts.
• SMBs can apply findings to role design, training, and measurement.

Highlights

• Contributes to understanding the relationship between AI task coverage, productivity, and workforce dynamics.
• Shows heterogeneous impacts across roles and sectors—measurement and training matter.
• Provides a framework to think about complementarity vs. substitution at the task level.
• Implication for SMBs: map tasks, not just roles, and redesign work accordingly.

Case study anecdote

A specialty manufacturer cataloged repetitive documentation tasks and piloted AI for draft creation and compliance checks. With training and QA, documentation time fell and engineers focused more on design iterations. The firm updated role expectations to include prompt ownership and review steps, which stabilized quality while throughput rose.

Guidance for SMBs

• Pair pilots with role clarity: list tasks to assist and tasks to keep human‑led. Train against both.
• Measure over time: throughput, error rates, rework, and satisfaction (customer/staff).
• Update job descriptions and review prompts quarterly as workflows evolve.
• Involve HR/ops to align incentives and recognition with new workflows.

Lessons & metrics

• Assisted documentation/reporting reduces time and variance.
• Training and QA are essential to sustain quality.
• Task‑level mapping clarifies where AI adds leverage and where human judgment is critical.

Learn more

Read the full article: Federal Reserve — Working Paper 2025-053

Looking for help to implement similar results? See our services.