Ethics Practices in AI Development: An Empirical Study Across Roles and Regions

Ethics knowledge varies significantly by role: The study of 414 AI professionals across 43 countries found that familiarity with AI ethics principles, government initiatives, and risk mitigation strategies differs substantially depending on job function—suggesting that ethics training cannot be one-size-fits-all and must be tailored to what different roles actually need to know and do.

Geographic differences matter: The inclusion of participants from 43 countries revealed regional variations in ethical perceptions and practices, indicating that AI ethics cannot be implemented uniformly worldwide and must account for different cultural, regulatory, and institutional contexts.

Collaboration across roles is essential: The research emphasizes that ethical decision-making in AI development requires diverse stakeholders—including managers, developers, quality assurance professionals, and security experts—to work together throughout the entire development process, rather than treating ethics as a separate concern.

Current practices are inconsistent: The mixed-methods analysis uncovered varying degrees of experience with risk mitigation strategies across the surveyed professionals, suggesting that many organizations lack systematic, standardized approaches to addressing ethical challenges in AI development.

Education and tailored solutions are needed: The study advocates for developing inclusive, role-specific educational strategies and solutions to promote ethics-aware practices, implying that current approaches to AI ethics training and implementation are insufficient for the diversity of people involved in building AI systems.