Large Language Models are trained to look at the widest range of information, for example, everything on the internet. That training enables it to produce completely new information. While it seems complicated from the outside, it is based on simple math and probability. LLM’s use a combination of the most probable results based on its training with some randomness to ensure that what it produces isn’t too repetitive. It uses a neural network much like the neurons in the brain to gather information and create.
Chat GPT is considered generative AI, because it uses the most information in the most complex way to come up with completely new works in text. Predictive AI simply reproduces what already exists based on more limited information. More like a remix or a forecast. Generative AI still needs humans to prevent results that could be overly biased or even dangerous.
Chat GPT is considered generative AI, because it uses the most information in the most complex way to come up with completely new works in text. Predictive AI simply reproduces what already exists based on more limited information. More like a remix or a forecast. Generative AI still needs humans to prevent results that could be overly biased or even dangerous.