Recently, a new AI model, “GPT-Next,” has started to create huge buzz in the tech world, according to the CEO of OpenAI Japan, Tadao Nagasaki. This model, expected to be released by the end of 2024, is claimed to be 100 times more powerful than GPT-4, marking a substantial leap in AI capabilities.
But this exceptional performance of GPT-Next is not because of some increase in computation resources; rather, it is due to increases in its architecture and learning methods. Development in GPT-Next, in essence, stems from two different projects: Strawberry and Orion.
Project Strawberry aims at enhancing reasoning, problem-solving, and language processing capabilities of the AI. This also introduces “System 2” thinking, greatly elaborating on reasoning and analyses, most notably in areas like mathematics and programming. Correspondingly, Project Orion is an effort toward extending those capabilities to a multimodal system able to handle text, images, and video and integrate them into a versatile tool for a number of applications.
These further steps, therefore, make GPT-Next more accessible and usable as an AI model, probably lowering barriers to entry for organizations wanting to use state-of-the-art AI technology. The improvements being made in the model are charting the way for major innovations in scientific research, health care, and business, among other areas, as this will offer deeper insights and more accurate predictions.
Nagasaki also made the point that, in addition to all this, the rapid growth of ChatGPT, which currently has over 200 million active users, is bound to continue as of August 2024. Its enterprise version, ChatGPT Enterprise, also finds wide applications in large companies.
