Page 219 - AI for Good Innovate for Impact

AI for Good Innovate for Impact

Limitations of Existing Solutions: Traditional multi-cluster management methods relying on
manual experience and simple rules lack real-time performance and effectiveness when facing
a multi-cluster environment with dynamically changing resource supply. They are unable to
quickly identify the optimal resource allocation according to the actual situation. Change 4.2-Climate

Benefits and Drawbacks of the AI-based Approach:
1) It can adapt to environmental changes 24/7. Based on factors such as the number of
running tasks, cluster computing power utilization, energy supply, and carbon emissions,
it can flexibly adjust the resource scheduling scheme for multiple clusters to achieve the
goal of green and low - carbon development.
2) It can perform performance modeling on computing tasks based on massive resource
monitoring and task operation data, etc., thereby improving the overall computing
performance.
3) It can use multi-modal generative large models to explain to humans the reasons behind
scheduling decisions.

The proposed use case emphasizes the role of AI in enabling innovative and green and
economical solutions for sustainable development. By proposing the requirements for
intelligent low-carbon management, this system will boost the construction of data collection
capabilities and low-carbon capabilities in industrial production. Additionally, by optimizing
the utilization of infrastructure, the system enables limited infrastructure to support more
tasks, indirectly reducing the need for new infrastructure construction and optimizing resource
investment. This system complies with the goal of responsible consumption and production by
optimizing the green utilization of resources throughout the AI lifecycle. It will promote green
energy consumption patterns in computing clusters such as data centers and reduce waste
and carbon emissions.

Use Case Status: The use case is part of a larger research project

Partners: N/A

2�2 Benefits of the use case

1) It can adapt to environmental changes 24/7. Based on factors such as the number of running
tasks, cluster computing power utilization, energy supply, and carbon emissions, it can flexibly
adjust the resource scheduling scheme for multiple clusters to achieve the goal of green and
low - carbon development.

2) It can do performance modelling on computing tasks based on massive resource monitoring
and task operation data, etc., thereby improving the overall computing performance.

3) It can use multi-modal generative large models to explain to humans the reasons behind
scheduling decisions.

The proposed use case emphasizes the role of AI in enabling innovative and green and
economical solutions for sustainable development. By proposing the requirements for
intelligent low-carbon management, this system will boost the construction of data collection
capabilities and low-carbon capabilities in industrial production. Additionally, by optimizing
the utilization of infrastructure, the system enables limited infrastructure to support more
tasks, indirectly reducing the need for new infrastructure construction and optimizing resource
investment. This system complies with the goal of responsible consumption and production by

183