"One way to encourage more environmentally responsible use of AI is for companies to take a staged approach. To begin with, there should be a rapid assessment of the green ‘maturity’ of their ongoing AI projects, to identify opportunities for improvement that will ‘green’ them up. This could include influencing AI developers and project leads to encourage the use of eco-friendlier AI design, such as choosing models and training methods that emit less CO2. Establishing eco-overlays over their development paradigms requires clear governance protocols that spell out the eco-sustainability principles, methodologies and benchmarks to which AI development must conform, especially to GenAI technologies. Establishing tangible metrics that can be utilized to serve as markers to track the ecological benefits that AI generates and refining AI operations in the cloud environment to minimize carbon footprints even within the limitations of current technologies are all important steps. But how can we help promote a culture of perpetual learning around developers and focus attention on the environmental cost of training AI systems? Through peer-to-peer networks of training, for inspiring knowledge exchange and best practices. Another way is for organizations to provide certifications—with industry-standard criteria—for energy, carbon and water footprints of AI solutions. This will ensure GenAI can be deployed sustainably throughout industries while encouraging wider adoption of green AI."
Shared by Mathieu Plourde, like and 1 save total
