Responsible AI emphasises human oversight, transparency, accountability, safety and appropriate use.
Bias can enter through data collection, labels, sampling, model design or deployment. Fairness requires examining whether performance differs across relevant groups or conditions.
AI systems may process sensitive personal, behavioural, biometric or location data. Data minimisation, security, consent and lawful use are important.
Generative AI can create convincing false text, images, audio and video. Verification of source, evidence and provenance is essential.
AI-generated and AI-assisted content raises questions about ownership, licensing, attribution, training data and permitted reuse. Applicable rules vary by jurisdiction and context.
AI can automate some tasks and change others while creating demand for new skills. Effects differ across occupations and sectors.
AI is often most useful as decision support, with humans defining goals, checking outputs and accepting responsibility.
Unequal access to devices, connectivity, data, computing and skills can create unequal benefits from AI.
AI systems can face data leakage, manipulated inputs, model misuse and insecure integrations. Security controls should match the risk.
AI governance uses policies, standards, risk management, documentation and accountability mechanisms. Laws and policies are evolving, so current official sources should be checked for jurisdiction-specific requirements.
Sustainable AI considers energy, computing resources, hardware lifecycle and environmental impact alongside social benefits.
You may practice now or skip directly to the next unit.