An AI model is the foundation from which ChatGPT, Gemini and other chatbot applications are created. Models are pattern recognition systems that are fine-tuned to answer questions and generate original content. See
AI training.
The AI programmer designs the model, which is a "neural network" with interconnections between each node called "parameters." These parameters start out with preset values and are modified in the training stages to generate better results. Large language models (LLMs) can have billions of parameters and small language models (SLMs) have hundreds of thousands or millions.
After the network is designed, the model is trained using examples in huge data sets extracted from every source on the Internet. During training, the parameters are adjusted to improve the results. For more details, see
deep learning and
neural network.