Safe AI for Everyone: A Plain-English Introduction
🔒 This course requires registration
To access this course and all our learning materials, please register for the AI Fluency programme.
Register Now →Frequently asked questions
What is an AI hallucination?
An AI hallucination is when a model generates false information that sounds confident and plausible. Large language models predict the next word based on patterns, so they can invent facts, citations, or events that never happened. Users must verify critical details against trusted sources to avoid spreading misinformation.
How do deepfakes work?
Deepfakes use generative adversarial networks to create realistic synthetic media. One network generates fake images or audio, while another detects flaws. This process allows creators to clone voices or faces with high accuracy, making it difficult to distinguish real content from fabricated material without forensic analysis.
How does AI bias affect decision making?
AI bias occurs when training data reflects historical prejudices, causing models to favour certain groups. For example, hiring algorithms might penalise candidates with gendered names if historical data shows fewer women were hired. Developers must audit datasets and test outputs to ensure fair and equitable results across different demographics.
What data does an AI chatbot store?
Most commercial AI platforms store conversation logs to improve model performance and safety. Users can often delete specific chats, but data may remain in backups for a set period. Always review the privacy policy to understand retention times and whether your inputs are used for training new models.
How can I verify information from an AI?
Cross-reference AI outputs with multiple independent, reputable sources. Check original documents, academic papers, or official government websites to confirm facts. Treat AI responses as a starting point for research rather than a definitive answer, especially for medical, legal, or financial advice.
