Artificial Intelligence, particularly Deep Learning, has demonstrated impressive outcomes across various sectors, including autonomous driving, manufacturing, and healthcare. Achieving high accuracy in these applications necessitates vast amounts of data, significant computational power, and intelligent neural network design. Consequently, this has led to the development of increasingly large models, involving billions of parameters. This trend contributes to a considerable environmental impact. It is essential for AI to adopt sustainable practices, much like other industries have done.
This webinar recording will start by diving into the cost of AI by answering the following questions;
- How does training an AI system impact the environment?
- What is the cost of using AI models in production and at scale?
- Is Big Data really the answer for good networks en vue of the waste they produce?
- Should we really be migrating all processes to the cloud?
- How can we make AI systems sustainable?
To address these questions, the various definitions of Green AI and their underlying principles will be presented. Subsequently, several strategies to reduce environmental impact will be introduced and examined. Topics will include transforming 'Big Data' into 'Good Data,' deployment on Edge devices, efficient architecture design, sustainable training, and optimization and inference techniques.
Ultimately, the talk will advocate for greater transparency regarding the costs associated with the AI Era and will call for aligning technological advancements with the necessity of environmental protection.
Speakers
