How is an AI model different from a RAG model?

bake_betty

In my understanding, a model is a black box that takes text as input and returns text as output, usually an answer to a question. A model can be trained, meaning it becomes smarter. Is it correct to think that a RAG agent is an agent that was not trained, compared to an LLM agent like GPT-OSS? Is a RAG system trained only by companies on their internal data? And is that the only difference compared to GPT-OSS? Please help me understand in simple...

Originally asked by voipp on Stack Exchange (ai) (CC BY-SA 4.0).

uphilltech

Great question — answering as the UphillTech team:

An AI model can mean different things, but I think you're referring to a machine learning (ML) model. There are many ML models, e.g. linear regression, support vector machines and neural networks (NNs), which are the most successful models nowadays. A large language model (LLM) is a big neural network. They are big because they have many trainable parameters. You can think of an ML model as a function. If you fix the parameters, you get 1 specific function. So, in general, yes, ML models need to be trained, because otherwise they just compute the function that happens to correspond to their initial parameters. You must adjust these parameters for your specific task. Training refers to the...

Full answer by nbro: https://ai.stackexchange.com/a/50632 (CC BY-SA 4.0).

Need it done for your business? UphillTech.