The world of technology is buzzing with a new word – ChatGPT. The platform was made available for public beta testing on December 1st. Within a week of its launch, ChatGPT has reached more than 1 million users. There is speculation that the platform could replace Google search and even more human in the future. Are you amazed and want to know how it works? Well, you are in the right place.
According to the company, the AI platform was trained using Reinforcement Learning from Human Feedback (RLHF). Put simply, the human AI trainers first delivered conversations in which they played both the AI assistant and the user. In addition, the company had given the trainers access to written sample suggestions that helped the trainers compose their responses.
In addition, OpenAI has developed a reward model for reinforcement learning. That means the company wanted to collect some comparative data for better response mechanisms. To collect such data, the company used the conversations between AT trainers and the chatbot. The company chose a written message in the mode, tested several alternative completions, and the AT trainers ranked these messages. Therefore, it is trained to follow an instruction in a prompt and give a detailed answer. Users can just type their request and the chatbot will respond to them. But then, how is it different from other artificial intelligence (AI) chatbots? According to the creators, unlike other AI chatbots, ChatGPT can answer follow-up questions, admit its mistakes, challenge false premises, and reject inappropriate requests.
According to OpenAI, ChatGPT is a free service, but only during research preview. This means that one can simply go to the OpenAI website and click the “Try ChatGPT” button to start using the platform. You can either login or use your OpenAI account to use ChatGPT. The company also provided a sample on the website for reference.
ChatGPT is the latest technology in the Generative Pre-Trained Transformer (GPT) family. To put it in simple terms, it is the latest AI text generation tool. But it’s not free from bugs or limitations. OpenAI admits on its website that ChatGPT sometimes writes answers that sound plausible but are wrong or nonsensical. Also, the model is often overly verbose, overusing certain phrases.
Get all technology news and updates on Live Mint. Download the Mint News app for daily market updates and live economic news.
More less
