- Sustainability
- Sustainability initiatives
- Social
- Responsible AI Initiatives
Responsible AI Initiatives
LY Corporation (the "Company") is committed to protecting user privacy and ensuring proper information management while widely utilizing various AI technologies in services and development, in line with its Basic Policy on AI Ethics. Specifically, for generative AI, the Company aims to become the most active user in Japan by building and enhancing the technical infrastructure and usage environment for employees. Through the use of generative AI to improve operational efficiency, the Company aims to expand its revenue in the medium to long term while providing more convenient services to users.
Generative AI Adoption Cycle
Leveraging its strengths, such as extensive Japanese data and numerous touchpoints with Japanese users, the Company is developing a "highly practical" environment to propel the cycle of adopting generative AI.
Generative AI Adoption Cycle — Diagram
1. Developing utilization foundations (usage environment)
As guiding principles to use AI safely while protecting user privacy, LY Corporation establishes guidelines and rules such as Basic Policy on AI Ethics. For details, please refer to"Responsible AI."
2. Building utilization foundations (development environment)
LY Corporation adopts a multi-vendor strategy meaning that it works to create an environment where employees can choose from a variety of generative AI options, collaborating with the best partner to build the technological foundation for utilizing generative AI. The Company has concluded a usage agreement on all of the API provided by OpenAI, L.L.C. LLMs (large language models) of Google Cloud and Amazon Web Services are also available.
3. Propelling business usage
Training on the risks of generative AI and prompt crafting techniques is mandatory for all employees. After completing the training and passing the subsequent test, employees can use the Company-exclusive conversational AI assistant "ChatAI." These efforts aim to establish a safe environment for employees to utilize diverse generative AI, boosting productivity.
4. Application in services
LY Corporation offers various features leveraging generative AI in services such as LINE and Yahoo! JAPAN. For details, please refer to "Applications of Generative AI in Various Services."
Related Links
Responsible AI
Measures to Establish AI Governance
In June 2021, LY Corporation established the Expert Panel on AI Ethics where the appropriate use of AI in consideration of societal needs (particularly its ethical aspects), is discussed continuously with external experts. In July 2022, the Board of Directors resolved to establish the Basic Policy on AI Ethics, which is the guiding principle to use AI safely while protecting user privacy.
The Basic Policy on AI Ethics consists of eight principles, including the "pursuit of fairness and impartiality," "ensuring safety and security," and "protection of privacy," to prevent unjust discrimination or inappropriate bias against specific attributes or individuals.
- Creation of a better future and contribution to mankind prompted by the diversity of information
- Realization of a peaceful and sustainable society
- Governance controls
- Pursuit of fairness and impartiality
- Pursuit of transparency and accountability
- Ensuring safety and security
- Protection of privacy
- AI human capital development
Furthermore, in October 2023, the AI Ethics Governance Department was established under the Generative AI Group as a dedicated organization to adapt with the rapidly changing developments in generative AI. The organization drives the use of generative AI across the entire company and also collaborates with the companies of the LY Corporation Group.
Related Links
Agile Governance with an Emphasis on Proactive and Protective Approaches
AI technology is recently evolving at a rapid pace, particularly around generative AI. The environment is also undergoing swift changes with the emergence of major tech companies and new enterprises, as well as the development of regulatory frameworks both in Japan and abroad. Under such circumstances, LY Corporation is creating a flexible agile governance structure that is not restricted by standardized operations and can accommodate these environmental changes.
Overview of Agile Governance Structure
The AI Ethics Governance Department establishes a consultation desk to address various risks associated with the use of generative AI.
Based on the nature of the inquiries, the Department flexibly establishes collaborative frameworks with relevant specialized divisions and provides advice and checks in accordance with the established guidelines and rules.
Furthermore, the Department closely monitors the status of Japanese and international initiatives, as well as trends among partner companies, users, and society, to make regular reviews of guidelines and rules as well as educational materials. It also supports the development of terms of use and policies for services. These efforts, along with ethical judgment criteria based on internal cases, are consulted with the Expert Panel on AI Ethics, which is composed of specialists with outstanding expertise.
Through these comprehensive initiatives, the Company establishes a system to prevent issues associated with the use of generative AI and provides services that users can utilize with a sense of security and safety.
Related Links
Risk Assessment on AI Usage
A risk-based approach on the use of AI is being discussed, especially in Europe. LY Corporation also refers to the EU AI regulations to assess risks and make judgements according to the business operations.
| Risk levels | Overview of risk levels | Policies |
|---|---|---|
| 1 | Businesses that can significantly impact human lives and well-being | Give top priority to considering AI risks |
| 2 | Businesses capable of significantly affecting everyday life | Thoroughly consider AI risks |
| 3 | Businesses that are unlikely to have a significant impact on everyday life | Address general AI risks |
LY Corporation prioritizes the following AI risk factors and conducts appropriate assessments to mitigate their impact.
| Risk factors | Description |
|---|---|
| Quality risks | Risks associated with AI not sustaining anticipated performance levels and the degradation of generative AI output quality e.g., Hallucinations, lack of consistency, data drifts, etc. |
| Legal/ethical risks | Risks of AI generating discriminatory language and generative AI producing illegal or harmful content that constitutes rights violations e.g., Inappropriate outputs due to the amplification of biases in training data, outputs that infringe on intellectual property rights such as patents and copyrights |
| Security risks | Risks of AI being attacked by malicious users or third parties during its training or inference phases e.g., Information leakage, adversarial input, etc. |
In addition to the abovementioned risk factors, LY Corporation conducts a comprehensive analysis that includes compliance, privacy, transparency, and safety, during risk assessment and decision-making. When AI is used internally or in its products, various departments such as legal, privacy, security, and intellectual property collaborate to perform risk assessments from a specialized viewpoint and make holistic implementation decisions. For high-risk cases, decisions are made by CxOs (such as the CDO, CISO) or cross-functional committees.
Data handling which is essential in AI utilization is performed based on the Basic Policy on Data Protection and LY Corporation Group's Cybersecurity Policy.
Monitoring System Centered Around the Division in Charge of AI Governance
Related Links
Employee Education
LY Corporation conducts mandatory training for all employees* to understand the risks associated with the use of generative AI, such as information leakage, infringement of rights, privacy violations, and hallucinations. This training consists of e-learning and tests, and it also covers techniques for crafting prompts to improve the quality of generative AI outputs. Employees must complete this training to use the company-exclusive conversational AI assistant, "ChatAI." The training content is regularly reviewed and updated in response to social conditions and advancements in AI technology, and all employees undergo the training approximately once every six months.
*All directors, permanent and contract employees are eligible.- Training completion rate for the second half of FY2023: 98%
| Major Risks | Information leakage, violation of rights, inaccurate output, discriminatory output, privacy |
|---|---|
| Prompts | How to improve output quality (Good and bad examples of prompts, sample templates) |
Applications of Generative AI in Various Services
LY Corporation offers various features leveraging generative AI in services such as LINE and Yahoo! JAPAN.
Application in Yahoo! JAPAN Search
In executing the growth strategy for FY2024 and beyond, which focuses on reinforcing media and search domains based on ID linkage, Yahoo! JAPAN Search implements improvements by releasing features that enhance convenience using generative AI, based on the diverse needs of users.
Related Links
- Growth Strategy, Integrated Report
- Yahoo! JAPAN Search Provides AI-Generated Summaries of Reviews on Over 80,000 Restaurant Search Results (Press release | Japanese only)
- Yahoo! JAPAN Search Provides AI-Generated Recommended Travel Plans (Press release | Japanese only)
- Yahoo! JAPAN Search Provides AI-Generated Summaries of Product Reviews in Search Results (Press release | Japanese only)
Application in LINE app
The communication app LINE has launched the LINE AI Assistant, a service that answers user questions and consultations in chat rooms using APIs from OpenAI and other sources. Similar to chatting with friends, it provides an AI experience that makes everyday life more convenient by helping with information search, generating ideas for work or homework, translating and summarizing, editing images, searching for recipes, and more.
Related Links
Application in Yahoo! JAPAN Knowledge Search
Yahoo! JAPAN Knowledge Search was launched as a Q&A service which provides a forum where users can ask various questions and seek advice, as well as share their knowledge and wisdom by providing answers. Having celebrated its 20th anniversary in April 2024, the platform sees approximately 54 million questions and answers posted annually. However, a significant challenge has been that about 10% of questions remained unanswered. To address this issue, AI has been introduced to automate responses, reducing the number of unanswered questions and increasing the variety of answers. In this way, Yahoo! JAPAN Knowledge Search continues to offer a high-satisfaction service that meets user needs by combining the strength of human knowledge and wisdom with the latest generative AI technology.
Related Links
- Yahoo! JAPAN Knowledge Search Provides Proprietary AI-Generated Answers Leveraging Past Best Answers (Press release | Japanese only)
- Yahoo! JAPAN Knowledge Search Integrates Claude 3 by Anthropic into AI Answer Feature with Generative Answer Responses (Press release | Japanese only)
- Yahoo! JAPAN Knowledge Search Launches Trial of AI Answer Feature with Generative AI Responses (Press release | Japanese only)
- Why Yahoo! JAPAN Knowledge Search, Celebrating 20 Years, Partnered with Generative AI (LY Corporation Story | Japanese only)
Measures to Mitigate GHG Emissions from the Use of Generative AI
As the use of generative AI and advanced technologies expands, the volume of data processing is rapidly increasing, raising new risks of greenhouse gas (GHG) emissions. In response to this challenge, the LY Corporation Group is committed to achieving carbon neutrality in its business activities by FY2030, aiming to reduce GHG for Scope 1 (direct emissions) and Scope 2 (indirect emissions) to net zero through the procurement of electricity derived fromrenewable energy with additionality, among other measures. Furthermore, the Group is also targeting "net zero" for total emissions, including Scope 3 (other indirect emissions) related to suppliers and the supply chain, by FY2050, in line with its Basic Policy on AI Ethics of realizing a peaceful and sustainable society.