Can AI be an alternative for South Korean companies to achieve climate competitiveness?

By JuneSeong Lee –

A booth has been set up in the exhibition hall for Samsung SDS to introduce its Cello square solution during the 2024 Global Supply Chain Innovation Expo

The ‘era of global boiling’ warning from UN Secretary General António Guterres came after July 2023 recorded as the hottest month in the past 120,000 years. In a rush to reduce greenhouse gas emissions, the most common offender behind climate change, some of South Korea’s largest conglomerates are turning to Artificial Intelligence (AI).

Samsung SDS recently announced its carbon neutral logistics goal by applying Optical Character Recognition (OCR) in the digital platform. Samsung Electronics is also racing its new product launch based on ‘SmartThings Energy’, an energy management service that functions with the ‘AI saving mode.’ Korea Telecom, likewise, is expanding energy management of urban buildings through ‘KT Giga Energy Manager’, in which the ‘e-Brain’, a self-developed AI analysis engine, works behind it.

However, AI experts and industry leaders are quite skeptical, as AI machine training and development uses more electricity than any other forms of computing, which emits more carbon. They warn that companies should deeply consider whether the AI model is valuable enough to tolerate the company’s power use and carbon emissions.

In South Korea, carbon emissions triggered by huge cargo transport from logistics companies, steadily increased since 1990. It showed the second highest increase rate of 5.5% among the country’s total emitted carbon of the energy sector. “The logistics process of transporting and delivering products does emit a significant amount of carbon”, says a Samsung SDS official who accepted an interview at the 2024 Global Supply Chain Innovation Expo held in COEX, Seoul. “Cello Square of Samsung SDS can manage carbon emissions by reducing the use of pallets, containers, and trucks through loading optimization.”

For logistics companies, reducing carbon requires specific analysis and calculation of carbon emission data in the supply chain. Samsung SDS has been collecting Scope 3 carbon emissions since launching ‘Cello Square’, a digital platform in October 2021. It manages the entire process of the supply chain as Optical Character Recognition (OCR) automatically extracts shipping document information, saves on the platform, and creates additional necessary documents. Carbon emissions are calculated from OCR documents of cargo departure, arrival, and cargo weight, and the platform provides personalized dashboards and reports of emitted carbon to customers. EcoTransIT, an outsourced solution is also used in the calculation process. For evaluation, Samsung SDS released that Cello Square reduced carbon emissions by 8.9% and 5% after improving the pallet loading rate by 11% and 22.2% in Europe and North America.

Still, South Korea’s Scope 3 carbon emissions, which are on the hands of 133 companies, are mostly derived from the use of sold products (53%, 800 million tCO2e), according to the Carbon Disclosure Project (CDP). The Korea Energy Agency further revealed that 20% of the nation’s total energy is consumed in urban buildings, which accounts for 23% of total greenhouse gas emissions.

In response, Samsung Electronics launched SmartThings Energy in February 2022, which provides energy monitoring and saving functions of home appliances. The ‘AI saving mode’ monitors the real-time usage, and automatically reduces power use before reaching the progressive range. The company released product experiment results of SmartThings Energy, which saved up to 60% in power usage for washing machines, 20% for air conditioners, and 35% for dryers. The platform is currently preparing its launch in more than 20 countries to reduce energy consumption of communities and achieve carbon neutrality.

Likewise, Korea Telecom (KT) has been operating KT Giga Energy Manager, an energy solution service since its launch in 2016. Utilizing the ‘e-Brain’, a self-developed AI analysis engine, it manages energy of urban buildings by controlling HVAC (Heating, Ventilation, & Air Conditioning) that accounts for nearly half of the building’s energy consumption.

Despite the efforts of South Korea’s major conglomerates using AI to reduce carbon emissions, AI experts and industrial leaders are raising questions on whether AI can reduce greenhouse gas and respond to climate change.

In an interview with Cho Nam-ho, a Generative AI Specialist and Engineer for Samsung SDS, he expressed concern that “the cost and resources required to use AI in industrial settings can trigger a climate crisis.” He went on saying, “training one AI model requires more electricity than 100 US households use in a year, and nearly 500 tons of carbon are emitted annually during the learning process.” Edward Park, Chief Strategy Officer (CSO) for OmniousAI who previously served as AI department head for GS Retail, also says “AI is using too much electricity in its learning process.” He added, “AI technologies used for corporations neither cause nor reduce greenhouse gas emissions.”

In a situation where the AI industry is growing rapidly, it is time for South Korean companies to find accurate methods to measure and manage carbon generated within its AI models, along with the efforts to reduce carbon emitted in their internal business. Bloomberg recently diagnosed that the expansion of AI industries such as ‘Chat GPT’ may increase carbon emissions explosively. At the same time, AI machine training is becoming more important to catch up with social change and demands. Yet, nothing is known on how much AI machines produce electricity and carbon in its development process. Unless the continuous efforts to develop comprehensive data on electricity usage across the AI ​​industry, South Korean companies using Artificial Intelligence will find it hard to disclose transparent data on electricity usage, carbon emissions and respond to climate change completely.

__________________________________