LGU+ AI Agent for On-Device Development
- LG Uplus and LG AI Research are joining forces to enhance the reliability of artificial intelligence (AI), focusing on creating "AI for the people."
- At MWC24 in Barcelona, Spain, LG Uplus introduced its 'Partner AI' strategy.
- The press conference featured key figures, including Lee Sang-yeop, CTO of LG Uplus, and Lee Hong-lak, CSAI of LG AI Research.
사람 중심 LG AI연구원과 협력…”안전하고 윤리적인 AI 서비스 개발”
Table of Contents
- 사람 중심 LG AI연구원과 협력…”안전하고 윤리적인 AI 서비스 개발”
- 온디바이스 sLM 개발해 엣지 컴퓨팅 고도화…올 하반기 상용화 추진
- 법적 리스크 자동 분석 에이전트 개발…AI 윤리 거버넌스 구축
- LG Uplus and LG AI Research: AI Development and Ethical Focus FAQs
- 1.What is the main goal of the partnership between LG Uplus and LG AI Research regarding AI?
- 2. What is ‘익시(ixi)’ and what makes it unique?
- 3. What is an on-device sLM, and how are LG Uplus and LG AI Research developing it?
- 4. what are the expected benefits of using on-device sLM in 익시(ixi)?
- 5.How is LG Uplus enhancing the security of its AI services, particularly 익시(ixi)?
- 6. what is LG Uplus and LG AI Research doing to address AI ethics?
- 7. What is the ‘Data Compliance Agent’, and how does it contribute to ethical AI development?
- 8.What is EXAONE, and how is it related to the on-device sLM development?
- 9. when is LG Uplus planning to commercialize the on-device sLM?
- 10. Where did LG Uplus introduce their ‘Partner AI’ strategy?
온디바이스 sLM 개발해 엣지 컴퓨팅 고도화…올 하반기 상용화 추진
법적 리스크 자동 분석 에이전트 개발…AI 윤리 거버넌스 구축
LG Uplus and LG AI Research are joining forces to enhance the reliability of artificial intelligence (AI), focusing on creating “AI for the people.”
At MWC24 in Barcelona, Spain, LG Uplus introduced its ‘Partner AI’ strategy. This was unveiled during a press conference held on the 5th (local time).
The press conference featured key figures, including Lee Sang-yeop, CTO of LG Uplus, and Lee Hong-lak, CSAI of LG AI Research.
LG Uplus and LG AI Research have set a clear goal: “AI that understands people,provides convenience,and maintains safety and trust.” A key offering is the ‘익시(ixi)’ AI agent service, which LG Uplus emphasizes is the first of its kind in the mobile communications industry, as data isn’t stored on a server.
Lee Sang-yeop, CTO, stated, “As we enter the era of AI, the mobile app ecosystem will evolve into a diverse AI agent ecosystem.” He added, “In this landscape, we must consider both ‘light’ and ‘shadow,’ focusing on safety and responsiveness to errors and fraud.”
‘온디바이스 AI 에이전트’ 개발…연내 상용화
LG Uplus and LG AI Research are advancing on-device AI technology by enhancing the existing CPU-based system to utilize a neural processing unit (NPU).They are developing an ‘on-device sLM (small Language Model).’ By applying the NPU-based on-device sLM to 익시(ixi), they aim to improve efficiency and performance, reducing power consumption to a quarter and CPU usage to a tenth of current levels.
The ‘high-performance sLM’ is crucial for on-device sLM. LG Uplus is optimizing communication services based on LG AI Research’s EXAONE, recognized for its performance in the global market. LG AI Research plans to integrate a model that delivers performance even with a reduced size of the EXAONE 3.5 model into 익시(ixi).
CTO Lee noted, “We are developing a 2.4B model on-device sLM that can be used on smartphones, based on the EXAONE model specialized in Korean, with plans for commercialization this year.”
CSAI Lee hong-lak added, “We plan to gradually apply it, starting with mobile devices launching with the progress completed and the hardware ready.”
The 익시(ixi) with on-device sLM is expected to provide more personalized services through enhanced computational power. This will enable 익시(ixi) to evolve into an ‘액스퍼트 AI’ service capable of handling tasks such as phone calls, summarization, scheduling, and restaurant reservations.
Security enhancements include the use of quantum cryptography and data outflow prevention. CTO lee stated, “Our quantum cryptography technology boasts the highest level of security in Korea with a physical unclonable function (PUF)-based KMS, meeting national defense-level security standards (EAL5+).” He also mentioned,”We will be the first globally to showcase a smartphone-embedded quantum-resistant cryptography PUF chip.”
The development of technology to block various AI-driven attacks and misuse is also underway. This includes ‘Anti-Deepfake’ technology to detect manipulated audio and video content, with ‘Anti-Voice’ slated for release first this year.
윤리적 AI 학습·운영으로 ‘사람 중심의 AI’ 구현
LG Uplus and LG AI Research are focusing on ‘AI ethics’ to ensure “AI for the people.” With the rapid advancement of AI, AI ethics is a key concern, aiming to prevent misuse, ensure trustworthiness and fairness, and provide guidelines for ethical technology application.
LG Uplus is pursuing organizational operations and compliance systems across LG Group to ensure ethical responsibility for AI technology.
LG AI Research is structuring AI ethics governance to oversee decision-making in AI research and development,ensuring no biased decisions are made. It also acts as a hub for establishing AI ethics-related policies across LG Group affiliates.
To create ethical AI, LG AI Research has developed a ‘Data Compliance agent’ that uses AI to automatically analyze legal risks in data.
While traditionally, humans had to individually verify the risks in data used for AI learning, LG AI Research’s technology allows for more accurate and cost-effective risk assessment. This enables LG Group to analyze and evaluate data that may contain sensitive information or legal risks, facilitating the development of more reliable AI services.
LG Uplus is also pursuing independent AI ethics enhancement measures. This includes automatically blocking the input of personal information when customers enter it intentionally, identifying and blocking sensitive information, and refusing to answer violent or inappropriate questions to ensure AI services are provided ethically.
CSAI Lee concluded,”LG AI Research will secure global-standard technology,from the latest AI research to model development,and create human-centered AI through this.” He added,”We will expand synergy with LG Uplus so that customers can use personalized AI services with confidence through 익시(ixi).”
Here are some frequently asked questions (faqs) and detailed answers based on the provided article about LG Uplus and LG AI Research’s AI initiatives.
LG Uplus and LG AI Research: AI Development and Ethical Focus FAQs
1.What is the main goal of the partnership between LG Uplus and LG AI Research regarding AI?
Answer: The primary goal of the collaboration between LG Uplus and LG AI Research is to enhance the reliability of Artificial Intelligence (AI), with a specific focus on creating “AI for the people.” This involves developing AI that understands people, provides convenience, and maintains safety and trust.
2. What is ‘익시(ixi)’ and what makes it unique?
Answer: ‘익시(ixi)’ is an AI agent service developed by LG Uplus. It’s claimed to be the first of its kind in the mobile communications industry because data isn’t stored on a server,emphasizing user privacy and data security.This on-device processing sets it apart from cloud-based AI agents.
3. What is an on-device sLM, and how are LG Uplus and LG AI Research developing it?
Answer: an on-device sLM (small Language Model) is a language model that runs directly on a device (like a smartphone) rather than relying on cloud servers. LG Uplus and LG AI Research are developing an on-device sLM by enhancing existing CPU-based systems to utilize a Neural Processing Unit (NPU).The NPU-based sLM applied to 익시(ixi) is intended to improve efficiency, performance, and reduce power consumption and CPU usage substantially. They are specifically working on a 2.4B model on-device sLM based on LG AI research’s EXAONE model, specialized in Korean.
4. what are the expected benefits of using on-device sLM in 익시(ixi)?
Answer: The on-device sLM in 익시(ixi) is expected to provide several benefits:
More Personalized Services: Enhanced computational power will enabling more individualized customer experiences.
Expert AI Capabilities: 익시(ixi) will evolve into an ‘expert AI’ service capable of handling tasks such as phone calls, summarization, scheduling, and restaurant reservations.
Improved Efficiency: Reduced power consumption and CPU usage.
5.How is LG Uplus enhancing the security of its AI services, particularly 익시(ixi)?
Answer: LG Uplus is implementing several security enhancements:
Quantum Cryptography: They are using quantum cryptography technology with a physical unclonable function (PUF)-based KMS, meeting national defense-level security standards (EAL5+).
Data Outflow Prevention: Measures are in place to prevent data from leaving the device.
Anti-Deepfake Technology: Developing technology to detect manipulated audio and video content, starting with ‘Anti-Voice’ technology.
Smartphone-embedded quantum-resistant cryptography PUF chip: First globally to showcase it.
6. what is LG Uplus and LG AI Research doing to address AI ethics?
Answer: LG Uplus and LG AI Research are actively focusing on AI ethics to ensure ”AI for the people.” their initiatives include:
AI Ethics Governance: LG AI research is structuring AI ethics governance to oversee decision-making in AI research and development, ensuring unbiased decisions. It also acts as a hub for establishing AI ethics-related policies across LG Group affiliates.
Data Compliance Agent: LG AI Research has developed a ‘Data Compliance Agent’ that uses AI to automatically analyze legal risks in data used for AI learning making risk assessment more accurate and cost-effective.
Independent AI Ethics Enhancement Measures: LG Uplus is implementing measures to automatically block the input of personal information, identify and block sensitive information, and refuse to answer violent or inappropriate questions.
7. What is the ‘Data Compliance Agent’, and how does it contribute to ethical AI development?
Answer: The ‘Data Compliance Agent’ is an AI-powered tool developed by LG AI Research that automatically analyzes legal risks in data used for AI training.This is a significant step as traditionally, humans had to manually verify these risks. The Data Compliance Agent allows for more accurate and cost-effective risk assessment, ensuring that the data used to develop AI services is compliant with legal and ethical standards protecting sensitive information and minimizing potential biases.
Answer: EXAONE is a large AI model developed by LG AI Research, known for its performance in the global market. LG uplus is optimizing dialog services based on EXAONE. LG AI Research plans to integrate a reduced-size version of the EXAONE 3.5 model into 익시(ixi)’s on-device sLM to maintain high performance while minimizing resource consumption. The on-device sLM is based on the EXAONE model specialized in Korean.
9. when is LG Uplus planning to commercialize the on-device sLM?
Answer: LG Uplus plans for commercialization of the 2.4B model on-device sLM this year (as was current at the time the article was written). The integration will be gradual, starting with mobile devices launching with the progress completed and the hardware ready.
10. Where did LG Uplus introduce their ‘Partner AI’ strategy?
Answer: LG Uplus introduced its ‘Partner AI’ strategy at MWC24 in Barcelona, Spain.
This comprehensive FAQ covers the key aspects of LG Uplus and LG AI Research’s AI initiatives, providing a clear understanding of their goals, technologies, and ethical considerations.