Beyond policy, what concrete AI evaluation methods will be developed?

Question

Accepted Answer

Singapore and Microsoft to Jointly Develop AI Evaluation Methods Beyond Policy Frameworks

Singapore's Infocomm Media Development Authority (IMDA) and Microsoft have entered into a partnership focused on advancing artificial intelligence (AI) safety and security, extending beyond policy development to encompass the joint creation of concrete evaluation methods. This collaboration aims to build robust benchmarks and tools to assess AI systems, thereby strengthening AI evaluation and fostering greater trust in advanced AI technologies.
Joint Development of Benchmarks and Tools

The partnership explicitly moves beyond the mere creation of policy frameworks to the practical development of shared resources for AI evaluation. This includes a commitment to jointly building benchmarks, tools, and other evaluation methods that can concretely advance the science of AI assessment. These efforts are designed to provide tangible means for evaluating the safety and security of AI systems.
Strengthening AI Evaluation Sciences

A core objective of this collaboration is to enhance the field of AI evaluation sciences. By combining IMDA's governmental insights with Microsoft's extensive technical and operational expertise, the partnership seeks to fortify the methods used to evaluate AI. This is crucial for addressing the emerging risks associated with advanced AI and ensuring their responsible deployment.
Facilitating Information Sharing and Research

Beyond specific evaluation tools, IMDA and Microsoft will also engage in technical and research collaborations focused on AI safety and security. Furthermore, the agreement includes provisions for facilitating information sharing between the two entities. This broader approach aims to build a comprehensive ecosystem that supports both innovation and the safe, reliable development and deployment of AI.

Reading history

Hide →

This history is cleared when the page is closed.

Beyond policy, what concrete AI evaluation methods will be developed?

Loading

Was this article helpful?

Helpful

Not helpful

Report an issue

1

0

0

Share with friends