OpenAI's Chief Technology Officer, Mira Murati, has clarified speculation about the company's highly advanced AI systems: In an interview, she explicitly stated that the AI models in OpenAI's research lab are currently no more capable than publicly available models.
She told Fortune magazine: "We have these powerful models in our lab, but they are not significantly more advanced than the models that are freely available to the public."
Her statement also implies that if the current internal systems are only marginally better than GPT-4, then the performance of OpenAI's publicly available AI models is unlikely to see significant improvement in the near future.
This also means that a significant enhancement in the performance of OpenAI's public models is unlikely in the near future. After model training, it takes several months of fine-tuning and safety testing before the models can be released to the market.
Critics argue that this confirms that the expansion of such AI technology has reached a bottleneck. In fact, since the release of GPT-4, there has been no significant progress in the fundamental capabilities of large language models, especially in terms of logic.
Microsoft founder Bill Gates also anticipates that the leap from GPT-4 to GPT-5 will be smaller compared to the substantial progress from GPT-2 to GPT-4.
However, OpenAI CEO Sam Altman and Microsoft CTO Kevin Scott continue to promise significant advancements, with Scott indicating that these will be realized this year.