In a move to bridge the widening gap between veteran craftsmanship and the next generation of industrial workers, Mitsubishi Heavy Industries, Ltd. (MHI) and AI startup Algomatic Co., Ltd. Have secured a top honor for their work in digitizing human intuition. The partnership saw MHI and Algomatic Win Second Place in the NEDO GENIAC‑PRIZE Program, a high-stakes competition designed to push generative AI out of the lab and onto the factory floor.
The winning project focuses on one of the most stubborn challenges in heavy industry: the “formalization of tacit knowledge.” For decades, the most critical skills in high-precision manufacturing have existed only in the muscle memory and “gut feeling” of master technicians—knowledge that is notoriously difficult to write into a manual or teach in a classroom. By leveraging generative AI, MHI and Algomatic have developed a system that can analyze the subtle differences between an expert and a novice, effectively translating embodied experience into actionable data.
The project was recognized by Japan’s New Energy and Industrial Technology Development Organization (NEDO), which organized the GENIAC-PRIZE to accelerate the real-world application of generative AI. Out of more than 200 entries, only 42 projects were awarded prizes during the final ceremony on March 24, 2026. The program distributed approximately 800 million yen across four key industrial themes, signaling a concerted effort by the Japanese government to maintain its manufacturing edge through artificial intelligence.
Cracking the Code of TIG Welding
To prove the technology, the team focused on Tungsten Inert Gas (TIG) welding, a high-precision method essential for the production of everything from massive energy plants to aerospace components and rockets. TIG welding is prized for its quality, but it is famously difficult to master. The difference between a perfect weld and a failure often comes down to microscopic adjustments in angle, speed, and heat—nuances that experienced welders perform instinctively.
The MHI-Algomatic approach replaces traditional, slow-motion apprenticeship with an AI-driven analysis loop. The process is deceptively simple: videos of both expert and novice welders are recorded and uploaded to an agent AI. The system then utilizes multiple analysis modules to pinpoint exactly where the novice’s technique diverges from the master’s. By extracting these differences, the AI can illustrate “embodied knowledge”—the physical movements and timing that experts cannot easily articulate.
This transformation of tacit knowledge into explicit knowledge allows for a more systematic accumulation of skill within a company. Instead of relying on the presence of a single master welder to train a cohort, the AI provides objective technical evaluations and real-time feedback, potentially shortening the learning curve for new technicians.
The Economic Imperative of Skill Transfer
From a market perspective, this project addresses a systemic risk facing global manufacturing: the “silver tsunami.” As a generation of highly skilled technicians reaches retirement age, companies face a catastrophic loss of institutional knowledge. When a master welder retires without a way to digitize their expertise, the company doesn’t just lose an employee; it loses a proprietary production capability.
MHI is integrating this AI capability into its broader “Innovative Total Optimization” (ITO) corporate strategy. The ITO framework is designed to halve lead times and radically improve business productivity through “overall optimization.” By reducing the time it takes to certify a new welder and improving the consistency of high-quality outputs, MHI aims to expand its domain capabilities and accelerate the delivery of complex infrastructure projects.
The GENIAC-PRIZE competition highlighted four critical areas where generative AI is expected to drive the most value in the Japanese economy:
| Theme | Primary Objective |
|---|---|
| Manufacturing Knowledge | Formalizing tacit knowledge and embodied skills. |
| Customer Support | Improving productivity and response accuracy. |
| Administrative Review | Streamlining government and corporate review tasks. |
| Risk Mitigation | Developing technologies to discover and reduce AI risks. |
Beyond the Prize: Toward Practical Application
While the second-place finish provides validation, the true measure of the project will be its scalability. The ability to analyze video and extract skill gaps is a blueprint that could be applied to any manual trade, from precision machining to complex assembly. If the AI can “see” the difference between a master and a novice in welding, it can theoretically do the same for any task involving physical dexterity and spatial awareness.
For MHI, the goal is now the transition from a prize-winning proposal to a standard operational tool. The company has indicated it will pursue the practical application of this technology to directly impact its production lines, aiming to create a sustainable pipeline of talent that is no longer dependent on the luck of a traditional mentorship match.
For more information on the program’s goals and the winners, the NEDO GENIAC-PRIZE official portal provides detailed documentation on the competition’s framework and the 42 selected projects.
MHI will continue to refine the AI modules as part of its ITO strategy, with further updates on the technology’s integration into its aerospace and energy divisions expected in upcoming corporate filings. Those interested in the intersection of AI and industrial productivity are encouraged to share their thoughts and experiences in the comments below.
