The conclusion first. Before turning a master craftsman’s know-how into data, the equipment designer must first fix the rated values and limit values in documented form. Data collected on-site without a baseline becomes a record that “the master did it this way,” but it cannot answer “why was that value permitted.” This is the author’s view.
1. Key News Fact Summary
The government’s 2027 budget proposal includes the items below (Document 1, government budget table item 1).
| Item | Detail | Budget |
|---|---|---|
| Physical AI Demonstration | 8 major fields including manufacturing, construction | KRW 0.5 trillion (KRW 2.6 trillion total) |
| Manufacturing Tacit Knowledge AI | Digitizing master craftsman know-how | KRW 0.3 trillion |
| Power/Water/Industrial Complex Support | KEPCO capital injection, underground transmission lines, etc. | KRW 2.1 trillion |
The Ministry of SMEs and Startups (MSS) budget table separately lists “SME Physical AI Demonstration: KRW 172.8 billion,” with the main content stating “KRW 172.8 billion newly allocated to support physical AI venture/startup demonstrations and AI transition for small manufacturing sites” (Document 1, MSS paragraph). The relationship between the two figures (whether it is a departmental allocation or included in the total) cannot be confirmed from the information provided.
One piece of industry news from the same period: Two Line Cloud signed an MOU with PIKAI for joint pursuit of physical-based AI and MSA² industrial AX (AI transformation), announcing plans to build an SaaS platform utilizing equipment data and jointly develop services such as predictive maintenance³ (Document 2, Two Line Cloud item).
2. Mechanism-Perspective Impact Analysis
The budget item’s name is “Digitizing Manufacturing Master Craftsman Know-How.” It makes clear that the object being digitized is human judgment. Human judgment is always made on the premise of equipment condition, and if that premise is not documented, the data loses its context.
“Utilizing equipment data” is the starting point of the project. However, the equipment-data problem the author has seen on-site is not a lack of sensors but the absence of baselines. When a vibration value rises and the normal upper limit exists nowhere in the drawings or specification sheets, the master craftsman judges by feel, and the AI learns that feel. A model that has learned a feel collapses when the equipment changes.
This article breaks the assumption that “AI transformation is the job of the IT department and the data team.” What the mechanical design team must hand over is not a 3D model but the following four numbers.
| Item | Value the Equipment Designer Must Document | Note |
|---|---|---|
| Rated Value | Design operating conditions (speed, load, temperature) | Check whether stated in specification sheet |
| Warning Limit | Value at which margin relative to rated value begins to be consumed | Together with safety-factor calculation basis |
| Stop Limit | Stop value before structure/bearing/drivetrain damage | Must match interlock setting value |
| Wear/Replacement Criterion | Judgment value for replacing consumable parts | Where the most master-craftsman know-how is hidden |
If this table is filled in, the master craftsman’s judgment can be coordinated as “where relative to the baseline the intervention was made.” If it is not filled in, the judgment is merely a list of points with no coordinates.
In the field, the counterargument arises that “the master craftsman’s feel is itself the baseline, and to document it, that feel must first be extracted.” This is valid. The order can also be reversed: setting a provisional baseline through master-craftsman interviews and then verifying it through design calculation. However, both approaches must ultimately be confirmed once more through design grounds (load, stress, life calculations). The safety factor is verified twice — through the master craftsman’s feel and through calculation.
3. Spec Comparison Table
| Item | Confirmation Status | Note |
|---|---|---|
| List of 8 Physical AI Demonstration Fields | Unconfirmed | Document 1 and Document 2 contain no specific field names |
| Budget Allocation Priority (by Equipment Group/Process) | Unconfirmed | Not specified in either document |
| Relationship Between MSS’s KRW 172.8 Billion and the KRW 300 Billion Total | Unconfirmed | Unclear whether departmental allocation or included in total |
| In-House Equipment Rated/Warning/Stop Limit Documentation Status | Confirmation Needed | Re-confirmation required after specification-sheet audit |
| Whether Interlock Setting Values Match Specification-Sheet Limits | Confirmation Needed | Re-confirmation required after field measurement |
One-Line Summary
What the manufacturing tacit-knowledge budget requires before digitization is prior documentation of the equipment side’s rated, warning, and stop limit values.
¹ Tacit Knowledge: Knowledge not documented, remaining embodied in an experienced person. ² MSA (Microservices Architecture): A method of structuring software as small independent service units. ³ Predictive Maintenance: A method of predicting failure timing from equipment condition data and performing maintenance accordingly.