Ethical Considerations in AI Flashcards

A company uses AI to infer employee mental health status from productivity data without disclosure. This violates which ethical principle?

Answer: Non-maleficence and privacy

Inferring sensitive health information without consent breaches privacy rights and risks causing harm (non-maleficence) through stigma or discriminatory action.

The 'right to be forgotten' in AI contexts most directly challenges which technical requirement?

Answer: Data deletion from trained models (machine unlearning)

Machine unlearning refers to the technical challenge of removing the influence of specific training data from a deployed model to honor deletion requests.

Which approach to AI governance uses a tiered, risk-proportionate regulatory framework?

Answer: Risk-based regulation (as in the EU AI Act)

Risk-based regulation applies stricter rules to higher-risk AI applications while allowing lighter oversight for low-risk uses, balancing innovation and protection.

An AI content moderation system over-removes posts from minority language communities. The best corrective measure is to:

Answer: Collect representative training data for underserved languages and retrain

Collecting representative data and retraining addresses the root cause — insufficient coverage of minority languages in the original training set.

Which principle states that AI developers and deployers must be identifiable and answerable for harms their systems cause?

Answer: Accountability

Accountability requires clear assignment of responsibility so that when AI causes harm, there are identifiable parties who can be held legally and ethically liable.

A generative AI model produces realistic fake images of real individuals. The primary ethical concern is:

Answer: Non-consensual synthetic media enabling defamation and identity harm (deepfakes)

Deepfakes can be weaponized for defamation, harassment, and fraud without the depicted person's consent, causing serious reputational and personal harm.

Which best practice ensures that ethical requirements are addressed throughout the AI development lifecycle rather than only at launch?

Answer: Ethics-by-design: embedding ethical review at every development phase

Ethics-by-design integrates ethical review into requirements, data collection, model development, and deployment stages, preventing issues from being discovered too late.