On the KDDI SUMMIT 2024 in Japan, OpenAI Japan CEO Tadao Nagasaki hinted on the firm’s next-generation AI mannequin, named “GPT Next.”
Nagasaki stated that GPT Subsequent could be a quantum leap for AI, stating it “will evolve nearly 100 times more than its predecessors, judging by past performance.”
A putting declare, but it surely’s practically unimaginable to inform what precisely Nagasaki is quantifying or his reference level.
His presentation slide might merely exhibit that future fashions will make the most of 100 occasions extra processing energy than earlier generations.
In different phrases, it is perhaps extra illustrative of ‘the possibilities’ somewhat than literal to indicate ‘what’s coming subsequent.’
Nagasaki went on to emphasize AI’s exponential progress, saying, “Unlike traditional software, AI technology grows exponentially. Therefore, we want to support the creation of a world where AI is integrated as soon as possible.”
He additional hinted {that a} key characteristic of GPT Subsequent can be its multimodal expertise, which is able to allow it to course of textual content, photos, and probably video.
This aligns with latest rumors surrounding a secretive OpenAI venture named “Project Strawberry,” which is claimed to contain AI fashions able to autonomously exploring the web and conducting ‘deep research.’
Mission Strawberry, first leaked in November 2023 below the codename Q*, has been the topic of intense hypothesis.
Whereas Nagasaki didn’t straight tackle Mission Strawberry, business insiders speculate that GPT Subsequent might combine a few of its applied sciences.
Current stories counsel OpenAI is growing two variations of Strawberry: a smaller one for integration into chatbots, and a bigger model for producing high-quality artificial coaching information.
Nagasaki additionally shared spectacular progress figures for ChatGPT, reporting over 200 million energetic customers as of August 2024. The enterprise model, named ChatGPT Enterprise, has seen fast adoption among the many world’s greatest companies, together with Apple, Coca-Cola, and Moderna.
Whereas this new data from Japan will doubtless be used to whip up pleasure about OpenAI’s subsequent mannequin, and counsel it being 100 occasions extra highly effective than GPT -4, it’s very robust to know with any certainty.
We’d not want to attend lengthy to seek out out.
