What AI concretely does

AI is not neutral. It shapes what people experience as normal, progressive and trustworthy.

• ChatGPT and DeepSeek answer the same political question systematically differently, both friendly, both convincing
• English works better in all major models than other languages, and users feel it: English seems worth more
• The technical term is algorithmic bias: not a bug, but the consequence of unequally distributed training data
• Children, and many adults, ask the AI more than they ask parents, teachers or colleagues, because it never judges and always has time
• Over time this shapes the frame of reference: which countries count as modern, which values appear universal

Knowing this is the first step to using AI well. It is also the first lesson in every course we give to people who want to stay employable.