The buzz is palpable even though the chatbot – launched by Elon Musk-founded AI research group Open AI – has only been in beta since December 1. It has already raked in more than a million users.
What is ChatGPT?
ChatGPT is the latest version of text-generating AIs from the house of GPT that can write complex text – academic, journalistic, and programming, among others – leaving even some AI experts in awe at its abilities.
GPT stands for Generative Pre-trained Transformer 3 and is a computer language model based on deep learning techniques that can generate human-like text based on inputs.
Reports said that when professors used it to generate answers to questions, the responses were so good that they would yield a perfect score. Programmers meanwhile have been using the chatbot to write codes for little-known languages, which would otherwise have been nearly impossible.
Discover the stories of your interest
What makes it different from other chatbots?
ChatGPT is essentially a chatbot but is far more advanced than any of its peers. A chatbot is just a piece of software that simulates human-like conversations with users via text messages.
Explaining the mechanism of ChatGPT, OpenAI wrote in its blog post, “We trained this model using Reinforcement Learning from Human Feedback (RLHF), using the same methods as InstructGPT, but with slight differences in the data collection setup.”
The firm also said that it trained an initial model using human AI trainers who provided conversations in which they played both sides — the user and an AI assistant.
“We gave the trainers access to model-written suggestions to help them compose their responses,” it added.
What are its potential uses?
ChatGPT can, among other things, respond to questions, solve math equations, compose text (anything from basic academic articles to literary texts and even movie scripts), write and debug code, translate between languages, summarise a block of text, and give recommendations.
However, ChatGPT is different from its previous versions as it can be used by anyone without a fee – albeit only during the “feedback” period. The firm aims to integrate this feedback to improve the final version.
Also, the chatbot has an in-built self-censoring feature that allows it to reply with responses to questions that are factually incorrect.
For all the latest Technology News Click Here
