Artificial Intelligence: ChatGPT is becoming more current for paying users

Artificial intelligence
ChatGPT is becoming more current for paying users

“GPT-4-Turbo” was trained with publicly available information until the end of 2023, says the developer company OpenAI. photo

© Richard Drew/AP/dpa

ChatGPT sparked the hype about artificial intelligence over a year ago. The developer company OpenAI has now presented innovations to the chatbot.

The latest version of the chatbot ChatGPT for paying customers has a much more up-to-date knowledge of the world. The fresh AI model “GPT-4-Turbo” was trained with publicly available information until the end of 2023, said the developer company OpenAI. Until now it ended in April 2023. OpenAI had promised that the chatbot would have access to increasingly up-to-date information. Initially his knowledge only lasted until autumn 2021.

ChatGPT sparked the hype about artificial intelligence over a year ago. Such AI chatbots are trained with enormous amounts of information and can formulate texts at the linguistic level of a human, write software code and summarize information.

The principle behind this is that they estimate, word by word, how a sentence should continue. One disadvantage: The software sometimes gives completely wrong answers, even if it was only based on correct information. Developers are working on setting reliable guardrails for the programs to prevent something like this from happening.

The latest ChatGPT version will also formulate less verbose texts that are intended to appear more colloquial, OpenAI announced. A text message reminder pre-formulated by the chatbot to friends to respond to an invitation to a birthday dinner reads briefly and objectively – just as a human would write it. Previously, in the comparison demonstrated by OpenAI, ChatGPT would have written a flowery text about three times longer, including the observation that it wouldn’t be the same without celebrating those invited.

dpa

source site-5