Survey reveals wide gaps in AI ethics practices across development roles and countries

A new empirical study surveying AI practitioners worldwide has uncovered substantial differences in how people working across AI development roles approach ethics — and how well they understand the frameworks meant to guide them.

Researchers conducted a mixed-methods investigation involving 414 professionals from 43 countries in various AI roles, including managers, analysts, developers, quality assurance specialists, and information security and privacy experts. Their findings highlight both the urgency of ethics-aware practices and the gaps that exist in real-world adoption.

What the research found#

The survey revealed uneven familiarity with AI ethics principles, government regulations, and risk-mitigation strategies. Comfort with these topics ranged significantly depending on job function, geographic location, and other demographic characteristics. Rather than a uniform understanding of ethical practice, the profession shows a patchwork of knowledge and application.

The researchers emphasize that these differences matter. A developer in one region may have very different training and resources than a counterpart elsewhere, leading to inconsistent approaches to ethical challenges that emerge across the AI development lifecycle.

Why this matters#

As AI systems grow more powerful and widespread, the stakes of ethical lapses — algorithmic bias, privacy violations, lack of transparency — increase accordingly. Without coordinated, informed practices, the industry risks compounding these harms at scale.

The path forward#

The researchers advocate for collaborative, stakeholder-inclusive approaches to ethics in AI development. Rather than imposing one-size-fits-all guidelines, they call for solutions tailored to different roles and regional contexts. This includes targeted educational strategies and research into how to build genuinely ethics-aware practices across the sector.

The study points toward a future where ethical decision-making is integrated throughout development, not treated as an afterthought or compliance checkbox.