What is ChatGPT? Before jumping to our topic, we can look at how ChatGPT introduces itself:
“ChatGPT is a large language model that was trained by OpenAI. It is designed to be able to
generate human-like text in response to prompts given to it, and it is capable of doing so in a
variety of different styles and formats. It is based on the GPT-3 (Generative Pretrained
Transformer 3) architecture, which is a type of machine learning model that uses deep learning
algorithms to generate text. ChatGPT is designed to be able to generate conversational text that is
similar to what a human might say in response to a given prompt, making it well-suited for
applications such as chatbots and natural language processing tasks.”
What do you think? This sounds like a natural conversation. In fact, ChatGPT is a new AI-driven
chatbot that can answer some questions and even write a paragraph of essays. When we search
on Google for an answer, the search engine tries to retrieve webpages with higher matching
scores to the questions. We need to check each webpage for a satisfactory answer. As humans
write the content of webpages, the answer can be a good candidate. Unlike Google search,
ChatGPT is trained with all webpage contents crawled from the internet. It learns the style of
answering a question. Meanwhile, we can consider the content of an answer as the
summarization of the major answers online. Hence, sometimes the answer can be wrong even
though it reads like someone confidently answers a question. It is still too early to say it can
answer questions with strong logic.
As we are a computer hardware company, let us start to “interview” ChatGPT with questions on
GPU graphics cards. From the answers, we believe in the future we can have a smart chatbot
online to do customer service since most answers sound reasonable.