Who Decides the Dangers of Artificial Intelligence?
Society shapes how we fear artificial intelligence risks based on institutions rather than pure danger. Furthermore, safety investments and risk definitions often favor wealthy nations over the global south.
In short: Society shapes how we fear artificial intelligence risks based on institutions rather than pure danger. Furthermore, safety investments and risk definitions often favor wealthy nations over the global south.
When new technology arrives, who gets to decide what counts as a dangerous failure?
What happened, in plain words
An analysis published in Jornal da USP discusses how societies choose which risks to fear and who is held responsible when things go wrong with artificial intelligence. The text explains that people who build computer models often define the risks themselves, missing problems faced by everyday workers and ordinary citizens. It notes that international safety databases mostly record events reported in wealthy regions, while the rest of the world mainly supplies raw cases. The article also highlights that technology companies spend the vast majority of their safety budgets on wealthy countries rather than the rest of the world. Finally, it points out that a research group in Brazil uses court decisions instead of news reports to spot different kinds of technology harms.
Key points
- Societies choose risks based on social rules Sociology research shows that people fear certain dangers because those fears confirm their social institutions, while using risk language to assign blame.
- Builders control the risk definitions When the creators of technology also explain its dangers, they leave out the perspectives of workers who label data and everyday people treated as test subjects.
- Safety money follows wealthy markets Internal documents from a major technology company revealed that the vast majority of its global anti-misinformation resources went to the United States, leaving very little for the rest of the world.
- Brazil explores different ways to track harm A university partnership in Brazil looks at actual court decisions rather than news reports, helping uncover hidden harms that standard international databases miss.
Terms explained
- Artificial Intelligence — Computer systems designed to perform tasks that usually require human thinking. Example: A program that recognizes faces in photographs.
- Taxonomy — A system used to group and name things based on shared traits. Example: Sorting books into categories like fiction, history, and science.
- Epistemological — Having to do with how we know things and create categories of knowledge. Example: Deciding whether a new discovery counts as scientific fact.
Why it matters
Understanding who defines technology failures helps citizens and lawmakers build fairer rules so that protection reaches every community, not just wealthy nations.
What we still don't know
This text is an opinion article presenting one author's perspective on technology governance and sociology, rather than a universal scientific consensus.
Based on reporting from Jornal da USP. This is an independent explainer, written in our own words with AI assistance; Jornal da USP has not reviewed or endorsed it. Read the original for the full details.