Your browser does not support JavaScript!

Advancements and Applications of AI Technology in Businesses and Public Sectors

GOOVER DAILY REPORT July 26, 2024
goover

TABLE OF CONTENTS

  1. Summary
  2. Technological Advancements in AI
  3. Corporate Applications of AI
  4. Public Sector AI Implementations
  5. AI Ecosystems and Industry Events
  6. Key Issues and Challenges in AI
  7. Conclusion

1. Summary

  • The report titled 'Advancements and Applications of AI Technology in Businesses and Public Sectors' provides an in-depth analysis of recent advancements in AI and their applications across various sectors. Key developments include Saltlux's AI solutions like 'Guber' and 'LuciaGPT,' Samsung’s robotic automation in semiconductor production, and Maum AI’s miniaturization of Large Language Models (LLMs) using the 'MAAL-Albatross' model. Public sector implementations highlighted include AI-based initiatives by the Digital Platform Government Committee and the Ministry of National Defense. Additionally, the report covers the AI EXPO KOREA 2024, emphasizing the significance of AI ecosystems through participation from over 300 companies and global exhibitors. Major findings underscore AI's role in boosting efficiency, productivity, and innovation while addressing ethical and legal challenges.

2. Technological Advancements in AI

  • 2-1. Recent Developments in AI

  • Saltlux has made significant advancements with its AI search engine 'Goover' and Large Language Model (LLM) 'LuciaGPT'. 'Goover' is an AI assistant capable of automating tasks such as information retrieval, summarization, and document creation. This system is based on the 'Lucia2' AI model, which uses twice as much training data as 'Lucia1' and provides more accurate answers to specialized queries. Saltlux has invested 40 billion KRW over the past three years to develop 'Goover' and has launched it in both the US and South Korea. Furthermore, Saltlux's 'LuciaGPT', revealed at the SAC 2023 conference, mitigates issues like hallucination and enhances data security, making it applicable across fields such as finance and law.

  • 2-2. Miniaturization of Large Language Models

  • Maum AI has achieved notable progress in the miniaturization of Large Language Models (LLMs). It developed the 'MAAL-Albatross' model, which operates 70 billion parameters using just two NVIDIA A100 GPUs. This was made possible through quantization techniques, compressing parameters to 4 bits while maintaining performance. This advancement allows for on-device AI applications, ensuring data privacy as models can operate within the company's internal network. Maum AI’s lightweight models offer tools like chatbots that help employees by providing accurate responses based on internal data without data leaks.

  • 2-3. Quantum Compression Techniques

  • Maum AI has also leveraged quantum compression techniques to further the capabilities of their LLMs. By employing such technologies, they have effectively reduced the size of these models, facilitating their deployment in low-power chips and on-premise environments. This development supports various applications, ensuring that LLMs can function efficiently without compromising performance. Consequently, Maum AI provides enterprises with solutions that enhance productivity while safeguarding sensitive information.

3. Corporate Applications of AI

  • 3-1. AI in Information Retrieval and Language Processing

  • Saltlux has developed the AI search engine 'Goover' based on their AI model 'Luxia2'. 'Goover' serves as a customized virtual assistant, automating tasks like searching, summarizing, organizing, and document creation. It uses doubled training data compared to 'Luxia1' and provides information tailored to user interests. Saltlux invested approximately 40 billion KRW over three years to establish a subsidiary in Silicon Valley and develop this service, achieving record sales of 30.7 billion KRW last year through their B2B and B2G businesses.

  • 3-2. Automation through Robotics

  • Samsung Electronics is integrating AI and robotics into semiconductor production processes. They plan to establish fully automated factories using intelligent robots from Hurim Robot that perform tasks in hazardous environments, aiming for complete automation by 2030. The project will enhance productivity and reduce safety hazards in the semiconductor manufacturing process.

  • 3-3. Custom Chatbot Solutions

  • Maum AI has developed an ultra-lightweight large language model (LLM) called 'MAAL-Albatross' that operates on minimal hardware. This model compresses parameters to 4 bits while maintaining high performance, ensuring data security within company networks. Maum AI provides a package of tools, including the 'Chatplay' tool, enabling companies to create custom chatbots that generate responses based on internal documents. This solution enhances employee productivity and maintains data privacy, making it highly reliable and efficient for business operations.

4. Public Sector AI Implementations

  • 4-1. AI in Public Services

  • The Digital Platform Government Committee, together with the Ministry of Science and ICT, has initiated various projects integrating AI into public services. These initiatives include AI labor inspectors and smart firefighting safety systems. In total, eight public service projects have been identified to receive AI integration, with a budget of 7.7 billion KRW allocated for this year.

  • 4-2. AI Support Systems for Labor Law Enforcement

  • Maum AI has signed an agreement with the Digital Platform Government Committee and the Ministry of Science and ICT to develop AI-based labor inspector support systems. This system aims to aid in managing labor law issues such as wage arrears and workplace harassment. The AI will assist by automatically drafting and analyzing statements and searching case law, thus enhancing the efficiency and accuracy of labor inspectors.

  • 4-3. Innovations in Defense Using AI

  • The Ministry of National Defense is undertaking several initiatives to enhance defense capabilities using AI technology. During the Defense Data AI Proliferation Week, AI solutions like the AI-based coastal surveillance system, VR flight training system, and military medical imaging support system were showcased. Additionally, the defense AI policy report discussed the development and deployment of Generative Defense AI, which specializes in translating military terms, answering queries on internal regulations, and summarizing documents specifically on the defense network. Efforts are underway to expand the involvement of AI in defense training and operations.

5. AI Ecosystems and Industry Events

  • 5-1. AI EXPO KOREA 2024 Highlights

  • AI EXPO KOREA 2024, held from May 1-3, 2024, at Seoul Coex Hall D, featured participation from 11 countries and over 300 companies and institutions, with a total of 450 booths. This event has been integral in fostering AI ecosystems and advancing AI industries since its inception in 2018. Highlights included the showcasing of various AI technologies, such as generative AI, large language models (LLMs), and AI-human interfaces. The event drew extensive attention, with over 32,526 visitors and buyers attending over three days last year.

  • 5-2. Global Participation and Exhibits

  • Key participants and sponsors included Supermicro, SeeLab, K2S, Maum AI, Woukyeong Information Technology, WEKA, and Hyosung Information Systems. These companies presented a range of technologies from generative AI, large-scale language models, machine learning, and deep learning to AI hardware, software, and infrastructure. Additionally, there were displays of conversational assistants like ChatGPT, AI humans, autonomous driving technologies, and AI applications in sectors such as healthcare, finance, industrial manufacturing, and retail services. The AI Elemental Technologies and AI Applications & Services sections of the expo underscored the importance of AI in various industries, demonstrating practical applications and innovations.

  • 5-3. Industry Conferences and Forums

  • Alongside the expo, significant conferences and forums were held to discuss the latest AI trends and governmental and institutional initiatives. These included the 2024 AI EXPO Conference, AI EXPO KOREA SUMMIT 2024, AI Business Convergence Development Conference (AI Tech 2024), and the AI Convergence Industry Forum. These events explored real-world applications, investment strategies, and provided insights into the development and integration of AI technologies across different sectors.

6. Key Issues and Challenges in AI

  • 6-1. AI Sovereignty and Global Competition

  • The rise of AI sovereignty or 'Sovereign AI' has become crucial as AI technology gains prominence. 'Sovereign AI' refers to an AI system that operates independently while adhering to national regulations, incorporating a country's language, culture, social context, and values. For example, Naver’s 'HyperCLOVA X' accurately identifies the disputed territory Dokdo as Korean territory, contrasting OpenAI's 'ChatGPT,' which identifies the area as disputed between Korea and Japan. Multiple countries, including India, Japan, the UK, Taiwan, and others, are actively pursuing the development of sovereign AI models to strengthen their AI sovereignty.

  • 6-2. Ethics and Legal Considerations

  • The ethical and legal implications of AI are a growing concern. The introduction of AI technologies into various sectors has raised questions about privacy, fairness, and accountability. Efforts are underway to ensure AI systems comply with legal standards and ethical guidelines to avoid unintended consequences and biases. Additionally, companies and governments are working to develop frameworks and regulations to manage the integration and impact of AI across different fields.

  • 6-3. Future AI Potential and Research Focus

  • While the provided documents did not explicitly focus on future AI potential and research focus, it is clear that significant attention is being given to the evolving capabilities of AI. Companies are enhancing specific AI models, such as small language models (SLMs), which offer cost-effective and efficient solutions for particular tasks. These models can significantly reduce operational costs and improve performance in targeted applications, indicating a trend towards more specialized and applicable AI research.

7. Conclusion

  • The report underscores the transformative potential of AI technologies through advancements by entities like Saltlux with 'Guber' and 'LuciaGPT,' and Maum AI’s work on LLM miniaturization. AI-driven automation, exemplified by Samsung’s efforts, reflects a significant shift towards increased productivity and safety in industries. Public sector initiatives by the Digital Platform Government Committee demonstrate AI’s capacity to enhance services like labor inspection and defense operations. However, challenges such as ethical considerations, legal frameworks, and the concept of 'Sovereign AI' require ongoing attention. Addressing these issues while fostering AI ecosystems, as showcased at AI EXPO KOREA 2024, will be essential for sustainable growth and practical applicability of AI technologies in real-world scenarios. Future prospects include further integration of AI in diverse sectors, focusing on developing robust, secure, and ethically sound AI systems.

8. Glossary

  • 8-1. Guber [Product]

  • An advanced AI solution by Saltlux focused on information retrieval and language processing, contributing significantly to enhancing corporate competitiveness.

  • 8-2. LuciaGPT [Product]

  • Another AI product by Saltlux, known for its innovations in language processing and making considerable impacts across various industries.

  • 8-3. Maum AI [Company]

  • A leading AI company involved in several important projects, including the labor law enforcement support system and leadership in miniaturization of large language models.

  • 8-4. AI EXPO KOREA 2024 [Event]

  • A major industry event highlighting the latest developments in AI technology, featuring participation from over 300 companies and various industry forums and conferences.

  • 8-5. Digital Platform Government Committee [Organization]

  • An agency involved in integrating AI solutions into public services, such as the labor law enforcement support system developed in collaboration with Maum AI.

9. Source Documents