The potential harms of AI

What are AI harms?

• Positive outcomes of AI tools.

• Negative consequences that can arise when artificial intelligence tools are used irresponsibly or designed poorly.

• Benefits of AI in educational settings.

• The process of developing AI models.

Which scenario best illustrates allocative harm?

• An AI-powered speech recognition system struggling to understand diverse accents.

• An AI tool generating educational content that only features boys in STEM fields.

• An AI-generated deepfake video of a school principal spreading misinformation.

• An AI-powered student support system incorrectly flagging a student as "low priority" for tutoring based on flawed data.

What is an example of Quality-of-Service harm?

• An AI system denying a student access to a scholarship.

• An AI tool creating content that reinforces gender stereotypes.

• A speech recognition technology struggling to accurately understand diverse accents or speech patterns from people with disabilities.

• An AI system causing widespread misinformation in a community.

How does representational harm manifest in AI tools?

• By denying individuals access to essential resources.

• By reinforcing the subordination or negative portrayal of social groups based on their identities.

• By performing unequally for different groups of people.

• By causing physical harm through its development or use.

Which situation exemplifies social system harm?

• An AI tool misusing private student information.

• An AI system that struggles to process speech from certain cultural contexts.

• An AI-generated deepfake video of a school official spreading misinformation and eroding community trust.

• An AI system preventing a student from accessing an online account vital for schoolwork.