A research company backed by the world's richest man, Elon Musk, who leads the electric car giant Tesla, OpenAI, announced that its latest chatbot, ChatGPT, has acquired one million users in less than a week. However, despite its claims of being able to converse with humans, OpenAI warned that the bot's answers may sometimes be "problematic" and even exhibit "biased" behavior. OpenAI hopes to "collect user feedback to help us improve the system's performance." ChatGPT is the latest product of a series of artificial intelligence research, which the company calls GPT, an abbreviation for Generative Pre-Trained Transformer, which refers to pre-trained generative transformers. The system was fine-tuned through conversations with human trainers based on the initial version. Although Musk is no longer a board member of OpenAI, he mentioned on Twitter that the system also learned by accessing Twitter data. Musk said that the system has suspended access to Twitter data.
The research company OpenAI, in which Tesla's CEO Elon Musk has invested, has released a new chatbot called ChatGPT, which has already gained a million users in less than a week. However, despite claims that it can converse with humans, OpenAI has also warned that the chatbot's responses may sometimes be "problematic" and even exhibit "biased" behavior.
OpenAI hopes to "collect user feedback to help us improve the system's work". ChatGPT is the latest product in a series of artificial intelligence developments, which the company calls GPT, short for Generative Pre-Trained Transformer. The system was developed through conversations with human trainers based on the initial version.
Although Musk is no longer on the OpenAI board, he stated on Twitter that the system also learned by accessing Twitter data. Musk said the system has stopped accessing Twitter data.
The chatbot has left a deep impression on many users. OpenAI CEO Sam Altman revealed in a tweet that people are very interested in AI chatbots: a million users have already used it in less than a week.
The chat format allows the AI robot to answer follow-up questions, admit mistakes, point out incorrect premises, and reject inappropriate requests. Mike Pearl, a journalist who tried the chatbot and reported on it for the technology news website Mashable, said it was difficult to provoke the chatbot to say offensive words.
However, OpenAI warns that the chatbot ChatGPT may sometimes give "seemingly reasonable but incorrect or absurd answers".
Nevertheless, the company also stated that it is precise because of the cautious training of the machine that it may refuse to answer questions that it could have answered correctly.