OpenAI’s GPT-4 Turbo receives stealth update for 2023
OpenAI has quietly updated GPT-4 Turbo’s training data to include information through December 2023, significantly extending its knowledge cutoff from April 2023. This enhancement aims to reduce model "laziness" by ensuring more complete task execution, positioning the advanced model far ahead of older versions like GPT-3.5, which relies on data from early 2022. The update’s legitimacy faces scrutiny due to mixed user experiences on developer forums, where some report the model still references older dates while others confirm access to recent data. Despite initial skepticism, the integration of recent information underscores the rapid pace of model refinement, even as the company simultaneously showcases advanced capabilities in video generation through Sora. This development is relevant to open_data as it highlights the critical importance of current, accessible datasets in maintaining AI utility and accuracy. As models increasingly rely on up-to-date information to function effectively, the lack of transparent public datasets creates barriers for developers and researchers. Ensuring that knowledge bases remain fresh and open is essential for fostering trust and enabling broader innovation in the AI ecosystem.
Source: cointelegraph.comPublished on 2024-02-22
Related news
- Google releases Gemma family of open-source AI models inspired by Gemini
- Meet Gemma: Google's new open-source AI model
- The open-source AI boom is built on Big Tech’s handouts. How long will it last?
- Google's Gemma: All you need to know about new open-source AI model
- Google debuts new 'Gemma' open-source AI models based on Gemini