Green AI

Program Description

According to the 2024 AI Index Report, the computational footprint of the state-of-the-art language models has seen a 7 orders of magnitude growth since 2017, while the cost to train these models in the cloud increased 5 orders of magnitude. As a result, the estimated carbon footprint of a state-of-the-art large language model is equivalent to 50-90 human years. This makes the current rate of increase in model parameters, datasets, and compute budget unsustainable. To achieve more sustainable progress in AI in the future, it is essential to invest in more resource-/energy-/cost-efficient solutions. This inaugural P1 program aims to fulfill this need by providing a forum for researchers from academia and industry with a focus on machine learning, artificial intelligence, systems, and computer architecture to collaboratively explore how to make AI’s computational and carbon footprint more transparent and improve the resource efficiency of AI applications and systems.

Green-AI in this program adheres to the broad definition in the literature and refers to the efforts to reduce the environmental impact of artificial intelligence (AI) development and deployment. It’s a movement that seeks to balance AI’s technological advancements with environmental sustainability, encompassing practices like developing efficient algorithms, improving hardware optimization, using renewable energy, and generally optimizing the overall resource usage across the life cycle of AI models.

This program will be focused on two themes: (1) improving the efficiency of algorithms towards designing Green-AI systems; and (2) exploring the use of renewable energy sources such as solar/wind to power AI infrastructure. Through this program, the anticipatation is to establishing a vibrant network of “Green-AI” researchers within Denmark, fostering innovative collaborations, and leading by example towards a more sustainable future where AI is not seen as a curse rather a blessing.