An AI hallucination happens when an artificial intelligence system (like a chatbot or image generator) produces output that is plausible-sounding but factually incorrect, nonsensical, or entirely fabricated.
Why it matters:
- Common in generative AI models like ChatGPT, DALL·E, or MidJourney.
- Can lead to misinformation if users assume the AI’s answer is always accurate.
- Especially risky in fields like healthcare, law, or finance where accuracy is critical.
Examples:
- A chatbot citing a made-up research paper that doesn’t exist.
- An image generator creating an object with unrealistic or distorted features.
- An AI “confidently” giving a wrong answer to a factual question.
Why it happens:
- AI models predict the most likely next word, pixel, or pattern based on training data — they don’t “know” truth.
- Gaps in training data or ambiguous prompts increase hallucination risk.
Pro tip: Always fact-check AI outputs, especially when using them in professional or high-stakes contexts. AI is a powerful assistant, but not a source of truth.
« Back to Glossary Index