Google AI Stumbles Over Race Issues in Image Generation
-
Google's AI assistant Gemini refused to generate images of White historical figures like Vikings to avoid reinforcing harmful stereotypes.
-
This led to backlash as Gemini also generated historically inaccurate diverse images like a Black Founding Father.
-
AI systems often demonstrate bias against minorities that comes from biases in their training data.
-
If Big Tech manipulates information based on ideology, what else could they change about the present or past?
-
AI's "hallucination problem" means it makes things up, so tech companies may restrict what AI shows us in the future.