How Do Computers Recognise Images?
An image is a grid of numbers. Recognition works by passing that grid through layers that detect progressively more complicated patterns, from edges up to objects.
Clear explanations of artificial intelligence, machine learning and the technology shaping everyday life.
An image is a grid of numbers. Recognition works by passing that grid through layers that detect progressively more complicated patterns, from edges up to objects.
Open source means the source code is published under a licence that lets others use, modify and share it. Most of the internet runs on software built that way, usually invisibly.
Recommenders mostly do not analyse the item. They find people who liked what you liked, then show you what those people liked next, and that produces both the accuracy and the tunnel vision.
Encryption turns data into something unreadable without a key. Modern systems use two different kinds at once, because each solves a problem the other cannot.
A search engine cannot search the live web. It searches a copy it built earlier, and ranking is a separate problem done with hundreds of signals rather than one formula.
Training runs are measured in months of full-power computing, and answering a question costs far more than a search. The reason is arithmetic done on a very large scale, plus the cooling to remove the heat.
Language models are trained to produce plausible text, not true text. Confident errors are not a glitch on top of that design; they are what the design produces when knowledge runs out.
Machine learning is not teaching a computer facts. It is adjusting millions of internal numbers until the model’s mistakes get smaller. Here is what that really means.
A quantum computer processes information using the rules of quantum physics, letting it weigh many possible answers at once instead of checking them one by one.