AI Ethics and Safe Use, Saltlux Consortium Takes Responsibility

As artificial intelligence (AI) rapidly advances, social anxiety and concerns about its side effects are growing. Following inappropriate responses and copyright infringement controversies, misuse cases such as fake news and deepfake content are surging, prompting efforts everywhere to create "trustworthy AI."

The European Union became the first in the world to pass AI regulatory legislation, and the United Nations unanimously adopted a resolution urging international consensus for safe AI. On the 21st, the Ministry of Science and ICT announced the "New Digital Order Establishment Promotion Plan," adopting AI safety, trust, and ethics as one of its core tasks.

AI company Saltlux announced on the 23rd that it was selected through a consortium for the new support project titled “Research on Data Characteristics for Social and Ethical Learning and Enhancement of Ethics in Generative AI Models,” part of the first 2024 Information and Communication and Broadcasting Technology Development Project and Standard Development Support Project.

This is a national project under the AI and data sector, specifically within next-generation generative AI technology development. The goal is to develop technologies that improve model performance to strengthen the social and ethical aspects of generative AI. Notably, research on AI’s social judgment is being attempted for the first time in the world, with a project budget of approximately 4.9 billion KRW.

The Saltlux consortium includes top domestic AI experts such as Professor Kim Hak-su from Konkuk University, Professor Koo Myung-wan from Sogang University, Professor Chun Hyun-deuk from Seoul National University, Professor Ko Young-joong from Sungkyunkwan University, Professor Kim Ki-eung from KAIST, Senior Researcher Kim San from the Korea Electronics Technology Institute, Executive Director Jang Du-seong from KT, and Research Director Kim Kyung-sun from Diquest. In addition to engineering researchers in AI and deep learning, multidisciplinary researchers from philosophy, humanities, and social sciences are participating, forming an optimal team to enhance the social and ethical aspects of generative AI.

The detailed objectives of the project are ▲establishing generative AI social and ethical guidelines that correspond to various universal characteristics and domain-specific features ▲developing core technologies for a sustainable and scalable comprehensive social and ethical response service platform ▲dissemination through public release and demonstration of research outputs. An open platform called Ethics-GenAIOps will be launched for use by any generative AI model and service developers, aiming to secure over one million users.

Saltlux Lab Head Kim Jae-eun said, "Following the nationwide research and development project ‘Exobrain,’ the entire consortium has united to once again achieve innovation in national AI technology through this project." He added, "As a company aiming for a world where humans and machines communicate knowledge, we will continue to strive for the establishment of a desirable digital order."