Beware of AI ethics imperialism

In the Moral Machine experiment, in which people answered ethical dilemmas faced by hypothetical autonomous vehicles, my collaborators and I wanted to capture opinions from across the world. So we translated the website to 10 languages, including Spanish, Portugese, Russian, Chinese, Japanese, Korean and Arabic. This, combined with the fortunate fact that the website went viral, gave us a treasure trove of data. When we did the data analysis, we had 40 million decisions by millions of people from 233 countries and territories.

Having data of this magnitude enabled us to conduct data analysis for each country separately. We discovered fascinating cross-cultural variations in people’s expectations of AI ethics, a phenomenon that had received very little quantitative attention previously.

For example, people in every single country thought the vehicle should prioritize the safety of children and younger people over older people. However, the intensity of this preference varied considerably. In Western countries, which score high on individualism, the preference was stronger. Said in another way, in Eastern, more collectivist countries, there was higher value placed on the lives of the elderly than in the West.

Countries also differed in the extent to which they prioritized pedestrians who were crossing the street legally over those who were jaywalking. In countries with a stronger rule of law—a measure of compliance with laws and confidence in government—, people were more likely to sacrifice jaywalkers to save legally-crossing pedestrians. You can explore those differences, and many others, by comparing pairs of countries on the Moral Machine website.

While specific to self-driving cars, these results highlight a potential problem with AI systems more generally. With most AI systems being developed in technologically advanced countries, the ethical values of these countries can be imposed on other countries through the behavior of the AI systems they sell to them. This can get particularly tricky when it comes to politically, culturally and religiously contentious domains. Whether this might lead to a new kind of AI-mediated ‘clash of civilizations’ remains to be seen.

References

  • Awad, E. et al. The Moral Machine experiment. Nature 563, 59–64 (2018).

  • Huntington, S. The clash of civilizations? Foreign Aff. 72, 22–49 (1993).

Previous
Previous

Consider Localism

Next
Next

Prevent algorithm capture