Japan's NEC On Generative AI's Vanguard

In July 2023, NEC launched its generative AI service, which includes licenses for the use of its LLMs, hardware, software and consulting services. The service includes dealing with information leaks and inaccuracies, and creating models of the knowledge accumulated within companies in order to develop systems that can use it effectively.

NEC's Japanese-language LLMs outperform the competition by 59% to 95% according to the Rakuda benchmark survey conducted by YuzuAI, a Tokyo-based organization that promotes collaborative development of the technology.

NEC's LLMs are also scalable and can be linked with specialized AI applications such as math, physics and industry-specific functions. Low power consumption allows edge deployment of the company's smaller LLMs, which can then be combined with cloud computing services.

This enables the automation of various operations by breaking down business processes into discrete tasks, autonomously deploying and linking AI models, and managing security and networks. NEC's LLMs can also be combined with the company's imaging and audio-sensing technology to enable the processing of real-world events.

Evaluations conducted by the US National Institute of Standards and Technology (NIST) show NEC to be the world's leading developer of biometric authentication technologies including face, iris and fingerprint recognition. The face recognition system used at NEC's head office in Tokyo was able to identify masked employees during Covid.

According to market research company Frost & Sullivan, NEC is the market leader in face recognition technology applied to airport immigration and boarding procedures, national ID and law enforcement. Overseas markets include Vietnam, India, Europe and the US.

The company also claims a competitive advantage in its extensive video recognition AI, which converts videos to information expressed in coherent sentences.

NEC finds use for this in fields as diverse as insurance, healthcare, manufacturing, construction and aviation. NEC ranks first in US patents and patents pending for video recognition.

In another example, similarity assessment and location prediction technology can detect disaster situations and their exact locations from a constant flow of images through cross-referencing with satellite and aerial imaging. When combined with an LLM, an ongoing disaster can be quickly described, improving the speed of emergency response.

With regard to corporate security, a specialized LLM utilizing NEC's cyberattack risk assessment technology can quickly respond to events and inquiries with diagnosis results and reports that a person who is not an expert can understand.

In a project with Tohoku University Hospital, the creation of electronic medical records and documents with an LLM trained in medical terminology can cut the time required to prepare those records and documents in half, reducing the workload of doctors. Local government is another area where time-consuming paperwork can be greatly reduced with LLMs.

Over the next three years, NEC management is targeting generative AI sales of 50 billion yen (US$335 million). That is less than 2% of the company's total sales, but the figure understates the likely impact on the company's business.

The introduction of generative AI and LLMs into business- and industry-specific solutions should lead to a meaningful upgrade of NEC's telecom services, social infrastructure and national security-related operations. It should also have a discernible impact on how people perceive AI.

Sign up for one of our free newsletters

• The Daily Report Start your day right with Asia Times' top stories
• AT Weekly Report A weekly roundup of Asia Times' most-read stories

Mark Anderson, CEO of Pattern Computer, an American AI and machine learning company headquartered in Friday Harbor, Washington, asks this question about ChatGPT:

“At a time when the world has gone gaga over a piece of software that does little more than reflect our language, it is worth asking what the next 10 years of AI might look like, long after the love affair – and pendular blowback – are over.”

The Office of the Director of National Intelligence and Central Intelligence Agency (CIA), in its IC OSINT Strategy 2024-2026 document, writes:

“GAI [generative artificial intelligence] can be a powerful tool to enable timely and insightful OSINT [open source intelligence] production, including by aiding the identification of common themes or patterns in underlying data and quickly summarizing large amounts of text. At the same time, OSINT tradecraft and training must be updated and refined to mitigate the potential risks of GAI, including inaccuracies and hallucinations.”

Generative AI blowback, inaccuracies and hallucinations can be limited by carefully managed source intelligence and thoroughly tested applications – i.e., the quality control that is essential to any precision manufacturing process, for which Japan is well known. It's precisely what NEC is now doing.

Follow this writer on
X: @ScottFo83517667

Already have an account? Sign i Sign up here to comment on Asia Times stories OR

Thank you for registering!

An account was already registered with this email. Please check your inbox for an authentication link.