NEC showcases customized generative AI solutions that unlock new business value. (Source – Shutterstock)
NEC’s AI model drives global economic growth and addresses the challenges of AI adoption. Its generative AI promises a major productivity revolution thanks to LLM, who is fluent in Japanese. In addition, the company’s AI solutions are reshaping businesses and boosting global economic growth.
Generative AI breakthrough technology not only transforms mundane tasks like data classification, but also creates unique experiences with the ability to generate human-like text, music, and digital art. While this innovation has sparked consumer interest and various stakeholders are exploring the implications of this form of AI for business and society, the underlying understanding is often limited.
Rapid advances in generative AI make this challenge even more acute. Evidence of that rapid evolution was his November 2022 release of ChatGPT, which was quickly superseded by OpenAI’s more advanced language model, GPT-4, just four months later. The technology marathon continues, and in May 2023 Google announced several AI-driven features such as Search Generative Experience and new language model PaLM 2.
The potential for these breakthroughs to drive global economic expansion is the true hallmark of this company. According to a recent study by McKinsey, generative AI could generate between US$2.6 trillion and up to US$4.4 trillion in economic value annually across 63 different use cases. This highlights the tremendous potential of such technology to serve as a key driver of global productivity.
AI solutions customized by NEC for the future
Adding to this momentum is the positive attitude of NEC Corporation, a Japanese multinational IT and electronics company. NEC is redefining the customer experience by creating customized generative AI solutions that adapt to technology changes and create new business value. From this month, NEC’s Generative AI service will start providing unique language model licenses, dedicated hardware and software, consulting services, etc. in Japan, and plans to gradually expand into new markets after FY2024.
NEC will start the “NEC Generative AI Advanced Customer Program” from July this year with the aim of promoting collaboration. In this initiative, we are collaborating with about 10 companies and universities, aiming to provide comprehensive support for the use of language models, from the development of language models to the software and organizational structure necessary for efficient use.
NEC has stepped up efforts to transform its business and established the “NEC Generative AI Hub” on July 1. The professional organization of researchers, consultants, digital trust specialists and other professionals aims to generate approximately US$346 million in revenue over the next three years. These AI-related ventures.
Despite the optimistic outlook, the application of generative AI presents its own challenges, especially regarding security, reliability, and applicability. Beyond translation and summarization tasks, companies need to leverage AI to leverage and apply their own company-specific knowledge internally.
The business impact of generative AI. (Source – Shutterstock)
In addressing these challenges, NEC has made significant progress. The company began using generative AI for internal purposes in May, and within two weeks successfully set up a secure and easy-to-use system. This system is currently used by approximately 20,000 users, and has greatly improved efficiency by reducing the time required to create source code, documents, and minutes.
Building an AI business ecosystem
At NEC, we are further improving the system in order to optimize tasks common to all employees, such as document proofreading, and specialized tasks such as software development. Insights gained from internal use of language model-related services can help achieve these goals.
NEC’s proprietary language model is notable for its excellent Japanese ability and compact structure, allowing it to run on standard GPU servers. The “NEC Generative AI Service Menu” offers a large number of customer support services that contribute to rapid and effective business transformation. NEC also provides software “NEC Generative AI Framework” that streamlines the integration of language models into customer workflows.
This AI service and “Microsoft Azure OpenAI Service” can be customized according to customer needs, such as a secure on-premises environment. These services are mainly hosted at the NEC Inzai Data Center to ensure safe and efficient operation of the language models.
NEC’s lightweight model, which boasts 13 billion parameters, is designed to operate smoothly in both cloud and on-premise environments. Excellent Japanese comprehension, as demonstrated in business-related knowledge and reading comprehension benchmark tests.
This model has already greatly enhanced internal operations, making tasks such as documentation generation and coding assistance more efficient. While most language models focus primarily on English, NEC’s model stands out as a high-performance Japanese model, adaptable to any industry and ready to accelerate business applications and improve productivity. it’s finished.
Features include:
- Japanese Proficiency: NEC’s LLMs perform well on the JGLUE benchmark, achieving 81.1% accuracy in question answering and 84.3% in reading comprehension, outperforming major competitors, leading across the industry indicates a possibility.
- Compact structure: Despite its high performance, NEC’s LLM requires only one GPU, making it efficient and cost-effective. It can also be rapidly built and deployed in on-premises environments to ensure the security of sensitive operations.
Evaluation results by Japanese comprehension benchmark JGLUE (according to NEC research)
NEC achieved this superior performance by leveraging high-quality data and computation time while keeping parameter sizes manageable. The company built this high-performance language model in about a month using his AI research supercomputer, the company’s largest.
In the future, NEC plans to use the base model to create company-specific language models using private data, and work to improve performance and promote practical application through the NEC Generative AI Hub.
