IBM Vice President: "AI Must Gain Trust to Accelerate Development... Effective Regulation and Openness Needed"

Christina Mongobery, Chair of IBM AI Ethics Committee

IBM emphasized the trust and safety of artificial intelligence (AI). Trustworthy technology is essential for gaining users' choice, which in turn accelerates the pace of development. Effective regulation and an open community were cited as conditions for enhancing AI reliability.

Christina Mongobery, Chairperson (Vice President) of the IBM AI Ethics Committee, stated this during a group interview at Korea IBM in Yeouido, Seoul, on the 23rd. Vice President Christina visited Korea to attend the 'AI Seoul Summit,' held over two days starting on the 21st. IBM also joined the 'Seoul AI Corporate Pledge,' which emphasizes AI safety, at this summit.

Christina Mongobery, Chairperson (Vice President) of IBM AI Ethics Committee
[Photo by Korea IBM]

Vice President Christina believes that both effective regulation and corporate responsibility are necessary for safe AI. Effective regulation means regulating based on use cases rather than the AI technology itself. She said, "Algorithms for personal credit and restaurant recommendations are not fundamentally different," adding, "Regulation should be based on the cases in which the technology is used."

She stressed the need for ongoing international discussions. Just as AI summits were held last year in the UK and this year in Korea, and will be held next year in France, the international community must continuously discuss AI safety. Christina said, "If various companies and governments approach this in a fragmented way, it is difficult to ensure safety," and added, "Continuous collaboration among players is extremely important."

She also called for responsibility from companies that deploy and use AI. As part of this, IBM established an AI Ethics Committee five years ago, the first in the industry. This reflects the priority given to safety and ethics from the AI development stage. It is an effort to understand the impact of new technologies, minimize side effects, and balance innovation. She said, "New technologies are adopted and used when they gain trust," and "Only then can technological innovation accelerate."

On the other hand, she emphasized openness. She believes that disclosing technologies such as data and AI models is necessary to effectively prevent potential risks. The involvement of diverse stakeholders in thoroughly reviewing AI models and identifying vulnerabilities in advance increases the likelihood of solutions.

To promote openness, IBM announced the open-sourcing of the AI model Granite at its annual event 'THINK' held in Boston, USA, on the 21st (local time). It was released on platforms such as Hugging Face and GitHub, where developers can share technology. IBM explained that by releasing its highest-performing model, it aims to create opportunities for improvement.

Recently, IBM also launched 'InstructLAB,' a community project in collaboration with Red Hat, a global solutions company it recently acquired. This community allows developers to advance open AI models and add their own data to build models tailored to specific fields or industries. Additionally, IBM is working to ensure transparency for private models by publicly documenting what data is used and how training is conducted.