Given how much noise is circulating right now around "GPT" I thought it might be helpful to provide a tiny primer on the topic. This won't be a deep dive into the technology, but hopefully we can help make you more comfortable with the terminology!
First, what is "GPT"?
GPT stands for Generative Pretrained Transformer, and the latest evolution is version 3, thus "GPT-3". GPT is known as a language model. Language models are developed by feeding huge amounts of training data (text) into a deep learning model which is designed to organize and then simulate human-like output, a process which is called Natural Language Processing (NLP). This natural language can be requested in different formats such as long form essays, condensed text, summaries, and more. You can also "teach" GPT-3 by providing it examples, which means this system grows the more it's used. Amazing!
Okay, so what's the difference between "GPT" and "ChatGPT"?
Simply put, ChatGPT is a variant of GPT-3 which has been specifically designed for chatbot operations. That is, its pre-trained model was weighted towards short conversational interactions, making it ideal for that specific use. ChatGPT isn't as powerful or extensible as GPT-3, but because it's more focused it also tends to be more efficient in processing requests and so faster to respond (also ideal for chatbots).
Hopefully that clarifies things when you hear those terms! Over the last couple of months and certainly over the next year, major technology companies like Microsoft will be feverishly working to integrate NLP's like GPT (which Microsoft has an exclusive license to) and ChatGPT into their products. Initially because it's buzzy and new, but also because there is a universe of possibilities for how GPT-3, ChatGPT, and the other NLP systems such as Bloom, Megatron-Turing NLG, and Rytr will change the face of language.
No Comments Yet
Let us know what you think