Japan picks Fujitsu for GENIAC large language model project

Japan’s New Energy and Industrial Technology Development Organization (NEDO) selected Fujitsu for the R&D project launched by the Ministry of Economy, Trade and Industry’s (METI) Generative AI Accelerator Challenge (GENIAC) initiative to enhance the country’s capability to develop generative AI. 

This is related to the Enhanced Infrastructures for Post-5G Information and Communication Systems/Development of post-5G information and communication systems.

Under this subsidy project, Fujitsu will start R&D on technology combining knowledge graphs with large language models to realise LLMs that enable logical reasoning.

Fujitsu has been conducting R&D of generative AI technologies for application in business operations. Last May 10, Fujitsu in cooperation with a team of researchers announced the release of Fugaku-LLM , a large language model with enhanced Japanese language capability, and plans to continue to strengthen the development of specialised LLMs for various industries and businesses.

As part of its R&D activities, Fujitsu has been focusing on the problem posed by hallucinations in LLMs, a phenomenon in which generative AI creates plausible but incorrect or unrelated output. 

Fujitsu aims to realise a new technology that strengthens the reliability of LLMs by combining LLMs with knowledge graphs, ultimately enabling the application of LLMs in operations that require high levels of compliance and explainability, including tort determination and basis search in the legal field, internal control and accounting audits in finance, and symptom search and diagnostic support in the medical field. 

Also, Fujitsu anticipates that the newly developed technology will enable the realisation of an LLM for logical reasoning that creates clear and comprehensible outputs.

Fujitsu plans to offer the new technology to the Japanese market at the end of fiscal 2024, as a step to further enhance its technologies to protect conversational AI from hallucinations and adversarial attacks announced in September 2023.

To address the task of reliability in the output of LLMs, Fujitsu started development of LLMs specialised in knowledge graph generation and inference, one of the knowledge processing technologies, to make LLMs logically infer answers according to the knowledge graph generated from natural language regulations and rules. 

Fujitsu plans to position this technology as the core technology of a generative AI trust technology that eliminates output instability and ensures accurate output in compliance with large and complex regulatory and internal regulations, a technique Fujitsu aims to realise in fiscal 2024 as part of its company-wide AI strategy.

As part of the GENIAC program, Fujitsu will develop two specialised LLMs — an LLM for knowledge graph generation, which converts natural language documents into knowledge graphs to form knowledge, and an LLM for knowledge graph inference that searches for relevant information on knowledge graphs for a given question, aggregates it logically, and answers it.

To efficiently develop two specialized LLMs in a limited development period, Fujitsu will first develop a pre-trained LLM that is common to both specialised LLMs. The advantage of this LLM development is that it enables LLMs to handle both documents in natural language and knowledge graphs by adding a bilingual corpus to the pre-learning data. 

Specialised LLMs are then developed simultaneously by carrying out instruction learning for knowledge graph generation for the first model, and instruction learning for knowledge graph inference for the second model.