OpenAI, an independent research body founded by the world's richest man Elon Musk and Sam Altman, launched a chatbot, ChatGPT on Wednesday, and it has peaked at over 1 million users in just a week.
The chatbot has taken over the internet by storm. This platform has made the internet curious with its human-like replies and prompt answers.
ChatGPT is a conversational dialogue model, a chatbot trained by artificial intelligence AI and machine learning. It understands and responds to the natural human language, 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 writing human-like text. Deep Learning is a machine learning method with three or more layers of a neural network, which attempts to simulate the behavior of a human brain, which then allows it to learn like humans.
How is it different from Siri or Alexa, who can also talk and respond, tell a joke or recite a poem? How is ChatGPT different from other AI models that are already available?
It's different because ChatGPT will remember your earlier conversations for context, it will admit its mistakes, challenge premises, and sometimes even decline to answer.
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. When it takes an action, it is classified into two categories - desirable or punishable. The desired action is rewarded while an undesired action is punished. AI learns what works and what wouldn't with this trial and error method.
OpenAI has also used humans as trainers of this AI. These trainers played both the roles of user and AI assistant through conversations. The training method can be a bit problematic, as it can mislead the model.
An ideal answer would depend on what the model knows, rather than what the human demonstrator knows. This exciting new thing could be a limitation on the internet.
If a user asks a complex question or doesn't phrase the question well, the bot can decline to answer. It will decline to answer if it is not an appropriate question.
E-commerce company Meesho's cofounder, Vidit Atrayee, took to LinkedIn to sound off this warning that until a few months ago, everyone thought AI could only take repetitive jobs, but with ChatGPT clearly not the case. For now, ChatGPT is free to use for the research period.
CEO Sam Altman has already said that the company will monetize the platform in the future. He said in a tweet that the service will always be free, the compute costs are eye-watering and will have to be monetized at some point.