skip to main content

What this guidance covers

This guidance addresses a central question: From the perspective of psychological science, what should organizations consider when bringing AI tools into the work of their employees, teams, and leaders? Importantly, this guidance also focuses on people: their judgment, behavior, and well-being.

Important note: This guidance does not address whether an organization should adopt AI. It addresses what organizations should consider when AI is introduced into people’s work. The evidence comes from research on AI and other automated systems, and from long-standing psychological research on how people make decisions, learn, work in teams, and perform at work.

APA Style leaf logo Cite This Report in APA Style
American Psychological Association. (2026, November 16). Deploying AI into workplaces: What leaders need to know. https://www.apa.org/pubs/reports/ai-workforce

How this guidance is organized

The guidance is organized around seven objectives. Each identifies something an AI deployment has to get right for the people whose work it changes.

1. Stakes: Match safeguards to the stakes of the deployment

  • Relevant psychological and behavioral concepts: Judgment under uncertainty, prospective thinking, signal detection, error detection and recovery, common-mode human failure
  • Strength of evidence: Well established for error detection, recovery, and the difficulty of forecasting rare high-consequence events. Emerging for correlated failure across an organization using the same AI system.

2. Trust and reliance: Help people develop and tailor their trust in AI tools over time

  • Relevant psychological and behavioral concepts: Trust and reliance, automation bias, algorithm aversion, confirmation bias, mental models, metacognition, processing fluency, psychological safety, working memory, and attentional capacity
  • Strength of evidence: Well established for the underlying phenomena. Emerging for how individuals behave with AI systems specifically.

3. Accountability: Make human oversight meaningful rather than a formality

  • Relevant psychological and behavioral concepts: Meaningful human agency, diffusion of responsibility, role clarity, moral responsibility gaps
  • Strength of evidence: Well established for role clarity, diffusion of responsibility, and their consequences. Emerging for attribution and moral responsibility in mixed human-AI work.

4. Job design: Redesign both work tasks and the job itself

  • Relevant psychological and behavioral concepts: Cognitive load, attention, situational awareness, hidden labor, job and work redesign, job demands-resources model, incentive alignment, change overload, job design
  • Strength of evidence: Well established for cognitive load, attention, and the effects of poorly designed job demand-resource balance. Emerging for the size and distribution of verification work in AI-assisted roles.

5. Expertise: Protect the ways people build deep knowledge

  • Relevant psychological and behavioral concepts: Skill acquisition and decay, desirable difficulty, elaborative processing, novice-expert differences, cognitive offloading, deskilling, and out-of-the-loop performance
  • Strength of evidence: Well established for skill acquisition, practice, and decay. Emerging for the specific effects of AI tools on professional skill formation and change.

6. Fair process: Introduce AI through a fair process

  • Relevant psychological and behavioral concepts: Procedural justice (voice, transparency, consistency, correctability), psychological safety, job insecurity and role ambiguity, psychological contract, organizational change, attribution and contribution
  • Strength of evidence: Well established for procedural justice, psychological safety, and the effects of insecurity and role ambiguity. Emerging for disclosure norms specific to AI use.

7. Monitoring: Respect psychological boundaries and set limits on monitoring

  • Relevant psychological and behavioral concepts: Privacy and information boundaries, autonomy and dignity, algorithmic control, electronic performance monitoring and surveillance, reactivity to measurement, psychological safety, procedural justice, validity of inferences about internal states
  • Strength of evidence: Well established for the effect of monitoring on behavior, perceived control, well-being, and performance. Emerging for the specific effects and validity of AI-enabled inferences about individuals’ internal states.

Young professional woman gesturing to colleague's laptop screen

How to use this resource

This guidance is not a compliance standard, an APA professional standard, or an evaluation of any product. Instead, use it to guide decisions and check your reasoning. Specifically:

  • Use the checklist in this document before a significant AI deployment and periodically afterward.
  • Discuss the questions in this document with people who understand the technology and the work it will affect.
  • Revisit the objectives in this document on a regular schedule and after major changes, such as a new system, a change in AI model, a major upgrade, or a reorganization.
  • Which objectives matter most will depend on the organization’s work, goals, and risks.
  • Record the important decisions, who owns them, and when they will be revisited.

Older professional woman using laptop in office

A note on the evidence behind this resource

This guidance rests on two kinds of evidence.

  • Well established: Some of the guidance draws on decades of research on how people work with automated systems and how organizations affect behavior. Much of this research predates current AI, but the underlying human behavior is well established. Examples of well-established areas of psychological and behavioral research referenced throughout this guidance include: trust and reliance on machines, automation bias, attention, out-of-the-loop unfamiliarity, cognitive load, procedural justice, psychological safety, models of job performance, team effectiveness, leadership and change implementation, the effects of job insecurity and role ambiguity, and workplace behavior.
  • Emerging: Other parts of the guidance address effects being observed with current AI systems. The research on these effects is newer, so the guidance treats the findings as provisional, not settled.

Rapid Expert Advisory Panel (REAP)

Tara Behrend, PhD

John Richard Butler II Endowed Professor
Michigan State University

Emily D. Campion, PhD

Associate Professor & Henry B. Tippie Fellow
University of Iowa Tippie College of Business

HeeSun Choi, PhD

Assistant Professor of Psychological Science
Texas Tech University

Joanie B. Connell, PhD

President
Flexible Work Solutions, Inc.

Nancy J. Cookie, PhD

Professor Emerita, Human Systems Engineering
Arizona State University

Alexis Fink, PhD

Director and Senior Research Scientist
Center for Effective Organizations
University of Southern California

Don Grant, PhD

National Advisor of Healthy Device Management
Newport Healthcare

Theodore L. Hayes, PhD

Research Psychologist

Louis Hickman, PhD

Assistant Professor of Management
Virginia Tech

Richard N. Landers, PhD

John P. Campbell Distinguished Professor of Industrial and Organizational Psychology
University of Minnesota Twin Cities

Brandon May, PhD

Assistant Professor of Forensic Psychology
Florida Institute of Technology

Fred Oswald, PhD

Professor, School of Education
University of California, Irvine

Emanuel Robinson, PhD

Director, Board on Human-Systems Integration
The National Academies of Sciences, Engineering, and Medicine

Ericka Rovira, PhD

Professor of Engineering Psychology
United States Military Academy

Mindy Shoss, PhD

Professor of Psychology
University of Central Florida

Sara B. W. Troutman, PhD

Managing Scientist, Human Factors
Exponent

Heng Xu, PhD

Professor of Information Technology, Analytics, and Operations
Mendoza College of Business
University of Notre Dame


APA staff

Dennis P. Stolle, JD, PhD

Executive Lead Psychologist, Applied Psychology

Mark Chan, PhD

Director, Applied Psychology

Elizabeth Deegan, MLS

Director, Science Programs for Knowledge, Expertise, and Innovation

Corbin Evans, JD

Deputy Chief, Operations & Strategy
Senior Advisor to the Deputy CEO

Ludmila Nunes, PhD

Senior Director, Science Knowledge, Expertise, and Innovation

Date created: November 2026