No comments yet.
This would be an AI trained on a combo of code, usage narratives, images and video. For example, the base OS code may not have a calculator app but having been trained on a recent windows version (its code, maybe some screen recordings of someone using calculator etc.), it could ‘generate a calculator interface’ (as could a a DALL-E-like product) and some code (using something like copilot) to respond with relevant images upon interaction with the image.