AI Image Generators and Bias

What is a common issue with AI image generators like MidJourney?

• They often create biased images.

• They are difficult to operate.

• They are too expensive to use.

• They cannot generate high-quality images.

What did the analysis by Leonardo Nicoletti and Dina Bass reveal about AI-generated images of different professions?

• There was no noticeable bias in the images.

• Higher-paying professions were equally represented by all skin tones.

• Lower-income jobs were predominantly associated with lighter skin tones.

• Higher-paying professions were dominated by lighter skin tones.

How do generative adversarial networks (GANs) function?

• They use a single model to generate images.

• They rely on human input for each image.

• They only work with text data.

• They consist of a generator and a discriminator that compete against each other.

What is a representational harm caused by AI systems?

• Creating images that are too realistic.

• Reinforcing harmful stereotypes and biases.

• Using too much computational power.

• Generating images too slowly.

What is one proposed solution to mitigate AI biases?

• Eliminating the use of AI image generators.

• Ignoring the biases as they are not significant.

• Using more representative data sets.

• Allowing AI to self-regulate without human intervention.