AI engineering glossary · Generative media multimodal

What is object hallucination?

Object hallucination: A failure of vision-language models in which the answer mentions objects, attributes or text that are not in the image. It arises from strong language priors overriding visual evidence and is measured with benchmarks such as POPE.

Level: intermediateTopic: Generative media multimodal

Object hallucination: an example

Shown an empty desk, a model describes 'a laptop and a coffee mug' because desks usually have them.

Related terms

More Generative media multimodal terms

Where you learn it

Object hallucination is part of the AI Engineering Program, which teaches it with notes, a hands-on lab and an exam. The first lesson is free.

Before you go

Take the cheat sheet with you

One free PDF: the key AI-engineering terms, formulas and Python idioms. No spam; unsubscribe any time.