Disclaimer: Copyright infringement not intended.

Context

• OpenAI’s ChatGPT, the artificial intelligence-powered chatbot that has gone viral, has crossed one million users in less than a week since it was officially made available to the public.

What is ChatGPT?

• ChatGPT is a natural language processing tool driven by AI technology that allows one to have human-like conversations and much more with a chatbot.
• The language model can answer questions, assist with tasks such as composing emails, essays and code. Usage is currently open to public free of charge because ChatGPT is in its research and feedback-collection phase.
• Basically, ChatGPT is a conversational dialogue model, a chatting robot, trained by artificial intelligence (AI) and machine learning. It understands and responds to natural human language and answers questions, and talks like you were to talk to humans.
• It gets its name from GPT or Generative Pre-Trained Transformer. It is a deep-learning language model that specializes in generating human-like written text. Deep Learning which is a machine learning method with three or more layers of a neural network attempts to simulate the behavior of a human brain, which then allows it to learn like humans.

Who made ChatGPT?

• ChatGPT was created by OpenAI, an AI and research company. The company launched ChatGPT on November 30, 2022. If the name of the company seems familiar, it is because OpenAI is also responsible for creating DALLE•2, a popular AI art generator, and Whisper, an automatic speech recognition system.

How is ChatGPT different than any other AI models that are already available?

• Well, it's different because ChatGPT will remember your earlier conversations for context, it will even admit its mistakes, challenge premises, and sometimes even decline to answer.

How does ChatGPT work?

• A user can start by going to OpenAI's website, click on the Try ChatGPT button, and start using.
• OpenAI has trained ChatGPT using a training method known as Reinforcement Learning from Human Feedback OR RLHF. It uses a reward AND punishment system to train AI. So, whenever it takes an action, it is classified into two categories - desirable or punishable. The desired action is rewarded whereas an undesired one is punished. With this trial-and-error method, AI then learns what works and what wouldn't.
• OpenAI has also used humans as trainers of this AI.  It is through conversations that these trainers played both the roles of a user and of an AI assistant. But this training method can be a bit problematic, as it can often mislead the model.
• An ideal answer would depend on what the model knows, rather than what the human demonstrator knows.. and this could be a limitation to this exciting new thing on the internet.
• So, if a user asks a complex question or does not phrase the question well, then the bot can decline to answer. It will also decline to answer if it's not an appropriate question.

For now, ChatGPT is free to use for the research period only.

https://www.thehindu.com/sci-tech/technology/the-workings-of-chatgpt-the-latest-natural-language-processing-tool/article66230152.ece

Download PDF