GenAI Images have many of their own problems and issues, and often those connected intellectual property and copyright law are centered. However, bias and stereotyping is rampant across many GenAI tools. Issues include:
- Trained datasets hold significant bias and sustain stereotypes (race, gender, body type, etc)
- Image stereotypes can be blatant or subtle
- Stereotypes can be in the background/periphery of the image, not always main subject
- Ethical safeguards tend to be manual and sometimes are reactive
- Jailbreaking as civic action (example)
- Deep fakes can be created by training GenAI tools on sample images
Additional Resources to Learn about Bias in Technology
