Vision-Language Models
AI systems that understand both images and text, so they can describe pictures or answer questions about them.
In Plain English
Vision-language models are trained on pairs of images and text descriptions, so they learn to connect what they see with words. They can answer questions like 'What's in this photo?' or 'Find pictures of people smiling,' without needing separate AI for images and separate AI for language. These models power tools like image search, accessibility features (describing photos for visually impaired users), and content moderation. They're useful because the real world rarely separates pictures from context—you usually care about both together.
💡Real-World Example
A real estate agent uses a vision-language model to list properties: she uploads photos of a kitchen, and the AI automatically writes descriptions like 'Stainless steel appliances, granite countertops, open shelving.' She can also ask it questions like 'How many windows are in each photo?' without switching tools.
Related Terms
What did you think of our explanation?