OpenAI Not Developing GPT-5, Focusing on Enhancing GPT-4 and Addressing Safety Issues
OpenAI CEO Sam Altman has confirmed that the company is not currently working on a GPT-5 language model, instead focusing on improving the capabilities of the existing GPT-4 model. Altman made this statement during a virtual event at MIT, where he also discussed the open letter calling for a pause in the development of large-scale language models.
Addressing Safety Concerns and Ensuring AI Models Align with Human Values
While acknowledging the importance of ensuring AI models are safe and aligned with human values, Altman pointed out that the open letter lacked technical nuance regarding where development should be paused. He stressed the significance of increasing safety rigor as AI capabilities become more serious and emphasized that moving with caution is crucial.
OpenAI’s Commitment to Transparency, Performance, and Safety
OpenAI plans to continue its transparency efforts but has not released information on GPT-4’s training data or architecture due to competitive and safety concerns surrounding large-scale language models. Altman’s comments indicate that OpenAI is taking a measured approach to language model development, prioritizing improvements in performance and addressing safety concerns.
