AI, Fairness and Fakes: Bias, Deepfakes and Misinformation
Can AI Be Unfair?
AI learns from data made by people. If the data is unbalanced or contains unfair ideas, the AI can copy them. This is called bias.
Example: if a photo-recognition system is trained mostly on pictures of one group of people, it may work badly on others. This is why designers must test AI on many different people and fix problems.
"Garbage in, garbage out" means that poor data gives poor results.
Fake Pictures, Voices and Videos
AI can now make pictures, voices and videos that look very real but are fake. Fake videos or audio made with AI are called deepfakes.
Fakes are not always harmful. Some are jokes or art. But they can be used to mislead people or to hurt someone, so we need to be careful.
How to Check What Is Real
- Stop and think before you share. Does it look too surprising, too shocking or too perfect?
- Look for clues: odd hands or fingers, strange writing, blurry edges, shadows that do not match.
- Check the source. Who posted it? Is it from a reliable news organisation?
- Look for the same story from several trusted sources.
- Ask a trusted adult.
Do Not Spread Fakes
If you are not sure something is true, do not share it. Sharing false information can cause real harm, even if you did not mean to.
Honest Labels
It is fair for people to say when something was made by AI, so others know what they are looking at.
AI and Society
AI will change how people work. Some tasks will be done by machines and new jobs will appear, but people will still be needed for caring, creativity and good judgement. Running large AI systems also uses a lot of energy, so people are working on making them more efficient.
Who Is Responsible?
People, not computers, decide how AI is built and used. That is why we all need to use it kindly and fairly.
Curriculum link: National Curriculum in England: Computing programmes of study (DfE) — Key Stage 2: be discerning in evaluating digital content; use technology safely, respectfully and responsibly