Generative AI & the senses
Computer vision is the field of AI that lets computers understand images and video, including recognizing objects, faces, text, and movement.
To a computer, a photo is a grid of numbers describing pixel colors. Computer vision turns that grid into meaning: there is a stop sign here, a pedestrian there, this is a cat. It is one of the oldest AI goals and one of the first areas where deep learning caused a breakthrough.
The big shift came in 2012. A deep neural network crushed the competition on ImageNet, a huge photo-recognition contest. Networks learned their own visual features, such as edges, textures, and parts, instead of relying on hand-written rules. On many tasks, accuracy now rivals or beats humans.
Computer vision runs in cars, phones, factories, hospitals, and shops. It also raises privacy concerns, especially facial recognition in public spaces. Modern multimodal models fold vision into language models, so you can ask questions about what a camera sees.
A self-checkout camera recognizes that the loose item on the scale is a bunch of bananas, before the customer has typed anything.