OpenAI's GPT-4 Revolutionises Japanese Language Processing: Triple Power, Top Proficiency, and Market Dominance Threat
OpenAI triples Japanese language processing power with its latest large language model, GPT-4. Translation and summarisation capabilities have been significantly improved, reducing output time by two-thirds.OpenAI's proficiency in Japanese language processing surpasses that of its Japanese competitors.
U.S. artificial intelligence company OpenAI is set to revolutionise the Japanese language processing capabilities of its core large language model. With triple the processing power, OpenAI aims to compete with similar services being developed by companies in Japan.
"Japan is an important market" stated Brad Lightcap, Chief Operating Officer of OpenAI, during a press conference held in Tokyo to mark the opening of the company's Japanese subsidiary. Lightcap also showcased a custom large language model optimised specifically for the Japanese language.
OpenAI is renowned for its ChatGPT generative AI platform. The latest iteration of its large language model, GPT-4, boasts significant improvements in Japanese language processing. Notably, translation and summarisation capabilities have been enhanced, and output time has been reduced by approximately two-thirds. The upgraded model is expected to be widely available to businesses and customers within the next few months.
The accelerated generation of answers through AI-based services can greatly enhance user interactions. Speak, a U.S. startup utilising OpenAI's custom open AI model, managed to streamline data processing and reduce costs by an impressive 47%.
Since the launch of ChatGPT in November 2022, OpenAI has consistently released new and improved versions of its large language models, focusing on enhancing response accuracy and security. In a recent announcement, the company revealed that its latest service, GPT-4 Turbo, will be made available to paid members.
While OpenAI's large language model was not initially optimised for the Japanese language, it has already demonstrated a high level of proficiency. In fact, a proficiency index of large language models published by Weights & Biases ranked OpenAI models as the top four performers in Japanese language processing.
The generative AI boom ignited by OpenAI has prompted major Japanese companies such as NEC, NTT, and SoftBank Group to venture into the development of Japanese-language generative AI. Domestic startups like Elyza, which originated from the University of Tokyo, have also entered the scene.
However, none of these competitors are currently on par with OpenAI. While Japanese companies claim that their large language models are equivalent to older OpenAI models in terms of language ability, they are perceived as lagging behind the cutting-edge advancements of OpenAI.
If OpenAI can create an AI system that is knowledgeable about Japanese systems, customs, laws, and business practices, it could attract even more domestic companies and consumers. Lightcap expressed his intention to collaborate with Japanese companies, but it is clear that OpenAI's advancements could pose a threat to the Japanese AI industry.
Meanwhile, companies in other countries are also catching up to OpenAI in the development of large language models that support multiple languages. Google's Gemini and U.S. startup Anthropic's Claude 3, released in March, are said to outperform GPT-4 in certain areas.
OpenAI triples Japanese language processing power with its latest large language model, GPT-4.
Translation and summarisation capabilities have been significantly improved, reducing output time by two-thirds.
OpenAI's proficiency in Japanese language processing surpasses that of its Japanese competitors.
Source: NIKKEI ASIA