Hallucination
In AI, a hallucination is when a model, like a chatbot, generates information that is incorrect, nonsensical, or not based on its training data or provided facts.
How it works
It occurs because the model statistically predicts plausible-sounding text rather than retrieving verified information. This is a critical challenge for reliability, especially when AI is used for factual tasks like research or customer support.
Examples
- 1.A chatbot inventing a fake historical event with convincing details.
- 2.An AI coding assistant generating a function that uses a non-existent library.
- 3.A customer service bot confidently providing incorrect store hours.