OpenAI fixed ChatGPT's “laziness” with a new update

OpenAI announced that the model GPT-4 Turbo Updated, it can “perform tasks such as generating code more accurately than previous models” and generally aims to reduce the “laziness” of ChatGPT's AI in responding to users' questions.

In a new post on the OpenAI blog, the company mentioned the release of a preview version of the updated GPT-4 Turbo model, which has increased accuracy in performing various tasks. However, OpenAI didn't specify exactly what it changed to its model with this update.

In the past some users have claimed that the ChatGPT AI has become lazy in their tasks and does not complete the required tasks perfectly, blaming it on the lack of GPT-4 update. However, OpenAI has updated the GPT-4 Turbo model, which was trained in 2023 based on the new data, and is only available as a preview. So people using GPT-4 which includes data before September 2021 may still experience the same laziness issues.

OpenAI also notes in the post that more than 70% of people using GPT-4 through their API have moved to GPT-4 Turbo because of the more up-to-date database. There will also be more updates to GPT-4 Turbo over the next few months, including public access to text-based image generation.

OpenAI has also launched smaller AI models called Embeddings. The startup describes embeddings as “sequences of numbers that represent concepts within content, such as natural language or codes.” These new models, which include the basic version text-embedding-3-small and a more powerful version called text-embedding-3-large, are available now.