Large language models are trained on the largest parts of language information as possible, such as everything available on the internet. It uses its training to generate new information. This can range from writing essays or poems, having conversations or even written code. This technology is being used to help produce apps and websites, construct movies and video games and even discover new medicine. These models use differentiated forms of communication through generative and predictive models. Some language models first focus on learning and predicting letters in the alphabet to generate written word or code. Others focus on scanning through the internet first, then moving towards tokens. Such practices like these require large amounts of human tuning to help produce a wide range of results.
