LG presented its “AI Ethics Principles” that serve as the basis for right action and value judgment of all LG employees who research and use AI. It is a significant step forward in the development of ESG management.
The principles were carefully selected based on LG’s management principles of “Creating Values for Customers” and “People-oriented Management” as well as core values for developing trustworthy AI. LG plans to establish a concrete implementation plan so that its AI Ethics Principles do not violate or regulate the autonomy of research and is further employed to develop ethical AI which adds more value to customers’ life beyond technology.
LG’s AI Ethics Principles consist of 5 core values: Humanity, Fairness, Safety, Accountability, and Transparency. Each of them is further described below:
■ Humanity
• LG AI provides benefits to humanity and society.
• LG AI does not violate human rights.
■ Fairness
• LG AI treats all people fairly while respecting diversity.
• LG AI avoids unfair discrimination based on individual characteristics.
■ Safety
• LG AI operates in a safe and robust manner.
• LG AI predicts and responds to potential risks.
■ Accountability
• LG AI clearly defines roles and responsibilities for those involved in development and use.
• LG AI strives to fulfil the responsibility to operate appropriately in line with the intention of the human.
■ Transparency
• LG AI is transparent with our customers to help them better understand and trust how AI works.
• LG AI follows principles and standards to ensure transparency in our algorithms and data.
As the hub of AI research for the entirety of LG Group, LG AI Research newly established “AI Ethical Inspection TF (Task Force)” in order to properly publicize the importance of the AI Ethics Principles and to review and implement measures to practically apply them in various tasks. The AI Ethical Inspection TF aims to train all the members of LG on the AI Ethics Principles, thereby allowing them to contemplate on AI technology and its social responsibility. The TF is also in charge of examining ethical problems that may occur in the course of AI research and development.
With regard to the presentation of the principles, the Chief of LG AI Research Kyunghoon Bae commented, “The valuation of AI will vary depending on how people regard and use AI. Through ‘LG AI Ethics Principles’ LG will continue our research on the coexistence of human and AI and create genuine customer values.”
By the end of this year, LG AI Research plans to create an “AI Ethics Working Group” in collaboration with 10 LG subsidiaries to discuss major AI ethics issues. AI Ethics Working Group will be comprised of members from LG Electronics, LG Display, LG Innotek, LG Chem, LG Energy Solution, LG Household & Health Care, LG Uplus, LG HelloVision, and LG CNS. The data and case studies accumulated by AI Ethical Inspection TF and AI Ethics Working Group will be utilized to create ethical guidelines for each and every field of AI that LG is currently involved by the year 2023.
Since its foundation at the end of 2020, LG AI Research has spared no efforts in researching trustworthy and ethical AI technologies for its customers. One of the most notable technologies is the “hate speech detection” technology designed to help customers avoid hateful and discriminatory statements when using customer service chatbots or other AI services. Started as the idea to resolve the issue of malicious comments which are the root cause of various social conflicts, it is currently under development to not only analyze the words but also the context of each sentence to filter out aggressive and biased information.
“Explainable AI (XAI)” is another ethical AI technology developed by LG AI Research. XAI is an AI technology that can help humans to understand how and why an AI arrived at a specific decision and what, if any, caused an error in the process. For example, during a vision inspection in an automated manufacturing process, the AI not only analyzes an image and determines whether a product is defective or not, but also explains to the operator why the product is determined to be defective. Therefore, it is a technology that is essential in developing expert AI that helps humans make decisions in fields of expertise that require trustworthiness and transparency, such as medicine, law and finance.
In order to develop an ethical AI that is trustworthy, transparent and accountable, LG AI Research is advancing XAI technologies and at the same time, collaborating with multiple LG affiliates to explore ways to utilize them in our daily lives. Utilizing XAI technology, LG AI Research is building a demand forecasting model in collaboration with LG Electronics and LG Household & Health Care, and developing a predictive model for customized immune chemotherapy treatment in collaboration with LG Chem. Furthermore, LG AI Research is working closely with Seoul National University Graduate School of AI and the University of Michigan on technologies associated with AI ethics to develop safer and more ethical AI technologies. LG AI Research will continue to exert our utmost efforts to present high-performance, trustworthy, and safe AI for customers.
▶LG AI Ethics Principles (Link)