*NEW RESEARCH* Managing the sustainable AI paradox

In trying to address their sustainability challenges while also exploring options for deploying AI, organisations need to face up to what we call the “sustainable AI paradox”. The AI models that crunch the data necessary to tackle sustainability problems and innovate service delivery also consume resources and generate carbon emissions due to their energy-intensive nature. Harnessing the power of AI, while ensuring its development and deployment aligns with sustainability goals, is a critical challenge facing both suppliers and tech user organisations.

Determining whether an application of AI is sustainable is highly nuanced and the level of carbon emissions is just one factor. An assessment of efficiency vs. effectiveness can be an important consideration—an AI solution may emit higher levels of carbon compared to a lower emission (but less efficient) alternative, for example, but be more effective.

This report looks at the sustainability challenges that suppliers and tech user organisations face when deploying AI, given the technology’s energy-intensive nature. It explores a range of potential responses to the “sustainable AI paradox”, as well as making the case for AI lifecycle environmental assessments that can identify areas where sustainability could be improved (and provide insight into unintended consequences).

SustainabilityViews subscribers can download Managing the AI paradox: Harnessing AI while hitting sustainability goals now. If you are not yet a subscriber, or are unsure if your company has a subscription, please contact Belinda Tewson to find out how you can access the research.

Posted by Craig Wentworth at '07:00' - Tagged: datacentres   energy efficiency   sustainable-by-design   power capping