Keynote: AI Ethics in Global Context
Track: Keynote | K1 |
Tuesday, October 29, 2024, 1:30pm – 2:30pm
Held in: Serra 2
Presenter:
David Harris - UC Berkeley
SPONSORED BY: LanguageLine Solutions
AI is all the rage, and in many ways people in the localization community have long had a head start in understanding its potential, watching for decades as automated translation tools steadily improved. As such, people in this profession are perhaps particularly well-suited to understand so many of the harms and risks associated with AI. In the same way that a bad translation in a critical application could cause serious consequences, an error on the part of an AI system operating in a high risk domain such as transportation or health care could be costly or even deadly. AI biases also can be incredibly damaging, which is particularly evident in areas where AI systems are used to make highly consequential decisions like ranking job candidates, scoring credit applications, or advertising housing opportunities. Biases can also be damaging in the representational sphere — where text, images, audio, and video produced by AI can reproduce damaging cultural stereotypes or produce anachronistic errors by overcompensating, as was the case with Google Gemini’s images of racially diverse Nazis. This talk explores the ethical problems associated with AI and suggests some tools that localization professionals can use to avoid them.