"In the Era of Physical AI, SMEs Must Act Quickly to Turn Data into Assets... Cooperatives Should Set Common Standards"

“SME Innovation Strategies in the Era of Physical AI” Forum
SMEs Need to Turn Manufacturing Data into Assets to Drive Innovation
Shared Framework for Cooperatives to Establish Common Standards and Demonstration Infrastructure
Calls for Shared Inf

It was noted that, to respond to the era of physical AI, small and medium-sized enterprises (SMEs) must first begin building up manufacturing data and turning it into an asset. The proposal is to establish a system for sharing responsibilities, centered on cooperatives, by creating common standards and infrastructure so SMEs can use their data and know-how with AI.

On October 21, the Korea Federation of SMEs (KBIZ) held a policy forum at Lotte Hotel Jeju under the theme “What Innovation Strategies Should SMEs Pursue in the Era of Physical AI?” to discuss solutions for SME growth.

In her keynote address, Park Youngsun, chair of the Ministry of Finance and Economy’s Strategic Economic Advisory Group, emphasized the need to innovate work and manufacturing operations in response to physical AI. “We need to build a ‘K-Ontology’ standard and platform that connects national and industry data,” Park said. An ontology is a technology that organizes the meaning of data and the relationships among data so AI can understand and process it. She added, “We need to gradually expand physical AI based on connected manufacturing data and equip ourselves with ‘AI factories’ that turn data into intelligence.”

“Companies need to have two factories: one that makes products and one that creates intelligence. It is difficult for SMEs to do this alone, so they need to develop industry-specific ‘vertical AI’ centered on cooperatives,” Park said. “We should invest R&D funding and other resources in KBIZ, and establish ways for cooperatives to collect data and turn it into an asset.”

The Korea Federation of SMEs (KBIZ) held a policy forum at Lotte Hotel Jeju on the 21st, together with the Ministry of Finance and Economy, on the theme, “What Innovation Strategies Should SMEs Pursue in the Era of Physical AI?” Attendees, including Chairman Kim Kimoon (sixth from left) and Park Youngsun, chair of the Ministry of Finance and Economy’s Strategic Economic Advisory Group (seventh from left), pose for a commemorative photo. Korea Federation of SMEs (KBIZ)

Manufacturing data must be turned into an asset so AI can use it, and standards must also be established. In a presentation titled “The Vision for Physical AI and Ways to Turn SME Data into Assets,” Kim Junha, dean of the GIST Graduate School of AI Policy and Strategy, said, “The key competitive advantage in the era of physical AI lies in process data and skilled know-how accumulated on manufacturing floors.” He added, “We need a system for sharing responsibilities in which individual companies accumulate data, cooperatives establish common standards and a shared basis for testing, and the government provides the infrastructure to support them.”

Kim defined data assetization as ▲ giving data a name (context), ▲ making it trustworthy (trust), and ▲ assigning it a value (value). “Data assetization means preserving discarded logs, photos, and work journals, giving the data a pedigree, enabling AI to trust and use it, and making it profitable,” he said, stressing, “You can still buy a robot three years from now, but you cannot buy data from the three years that have passed.” He continued, “What physical AI most wants to learn is tacit knowledge, and the government also plans to begin developing AI solutions that use tacit manufacturing knowledge. We need to increase opportunities for participation, including by forming consortia with AI companies.”

“SMEs are slow to adopt AI because it is difficult for them to bear the costs and risks of failure alone,” said Jo Juhyeon, president of the Korea Small Business Institute (KOSBI). “We need a ‘collaborative innovation model’ that allows companies to share common foundations such as data and testing facilities while protecting each company’s competitiveness.”

A panel discussion on ways to expand the use of physical AI among SMEs also highlighted the need to build shared infrastructure, centered on cooperatives, to bring AI into manufacturing operations.

“Small businesses in the root industries have abundant manufacturing data and many operational problems to solve, but lack AI talent and investment capacity,” said Jang Yonghwan, chairman of the Gyeongin Foundry Industrial Complex Business Cooperative. “We need to expand SME participation in government research and development and demonstration projects, and provide support for long-term testing and operation in actual factories.”

“Simply providing support to purchase robots or solutions will not be enough to establish physical AI in the field,” said Oh Minseok, vice president of Megazone. “We need shared infrastructure tailored to each region and industry that can support everything from data preparation to on-site implementation and maintenance.”

“Physical AI is ‘manufacturing intelligence’ that makes integrated decisions about and operates an entire factory,” said Jang Youngjae, head of the Manufacturing Physical AI Research Institute at the Korea Advanced Institute of Science and Technology (KAIST). He proposed, “Let’s create a ‘factory manager AI for everyone’ model that starts with small-scale demonstrations and expands validated AI models across industries.”

Before the forum, KBIZ Chairman Kim Kimoon presented Park with “The AI Transformation of SMEs: Voices from the Field and Policy Tasks,” a summary of policy needs based on challenges SMEs face on the ground. Kim stressed that views from the field should be actively reflected in the development of the national AI strategy and SME AX policies.

“We should design step-by-step support policies suited to SMEs’ capabilities so they can participate as drivers of the AI transformation, and rapidly expand AI with industry-specific cooperatives at the center,” KBIZ Chairman Kim Kimoon said.

This content was produced with the assistance of AI translation services.