AI Dictionary › AI Fundamentals

AI Bias

Bias (nei sistemi AI)

Bias in AI systems is a model's systematic tendency to produce distorted or discriminatory results toward certain groups, topics or perspectives. Bias doesn't come from the model's "bad intentions": it is absorbed from training data, which reflects the prejudices present in human text, and from design choices.

Definition

Concrete examples: a CV screening system that penalizes female names because it was trained on unbalanced hiring history; an image generator that associates certain professions with a single gender; a language model that treats identical requests differently depending on the language they're phrased in.

For professional use, especially in HR, credit, healthcare and justice, bias is a legal risk as well as an ethical one: the EU AI Act classifies these uses as "high risk" precisely for this reason. Mitigation: curated datasets, fairness testing, human oversight on decisions that matter.

Related terms

• Training Data
• EU AI Act
• AI Hallucination
• Explainability (XAI)
• Machine Learning

More in AI Fundamentals

• LLM (Large Language Model)
• Token
• Context Window
• Temperature
• AI Hallucination
• Fine-tuning
• RAG (Retrieval Augmented Generation)
• Embedding

Put it into practice

• Real scenario: apply this concept with an instant AI verdict.
• Daily gym: short drills to train dictionary concepts.
• Playoff game: challenge other players, elimination-style.

Say what you mean. Get what you need.

Grace Certified, the AI coach that trains and certifies your prompt engineering, by Agora Intelligence.