AI will fall short of enabling sustainability, unless it’s sustainable by design.
“Technology, and AI in particular, has great potential to help our generation overcome key challenges such as global warming, delivering responsible innovation”, said Uljan Sharka, CEO and Founder of iGenius.
Studies have shown, for example, that applied AI can play a key role in enabling the achievement of some of the Sustainable Development Goals identified by the United Nations, such as more affordable and clean energy.
“Innovation and technology can be powerful enablers”, continued Sharka, “but they also come with their own price tag in terms of resources and overall societal, environmental and efficiency impact.”
Energy used in AI model training amounts to about 626,000 pounds of carbon dioxide, research says — that’s five times the lifetime emissions of an average US car.
AI sustainable by design
“How can we make AI that is in itself sustainable?”, continued Sharka. “Surely, the same values at the core of producing and supplying clean power, for example, can be applied to the development, maintenance and delivery of AI products.”
“We believe that in order to be sustainable, AI needs to be efficient but also accessible, simple to use. These principles are also the foundation of our AI advisor for data intelligence, crystal.”
Sustainable AI rests on three pillars for us:
This means reduced consumption of energy and technical resources.
crystal can make time-to-value 15x faster
crystal is not an additional tool that needs to be bolted on any other already complex analytics systems. Companies don’t have to make room for it, nor teach their employees about complex set-up phases, or provide them with training on how to use it.
It is an overarching layer that connects to and finds patterns between a company’s analytics systems already in place, making time-to-value 15 times faster.
Sustainability should extend well beyond the adoption phase.
crystal can increase adoption rates by 90%
Adoption of analytics currently is limited to the realm of digital-native or data literate employees. This is due to hurdles such as data complexity and scarcity of data literacy skills.
Making data accessible to anyone and simplifying the adoption process by adapting to user goals and expectations rather than yielding to data complexity, crystal can make adoption rates go up to 90%.
Reducing computing power in terms of consumption of energy and space.
crystal works efficiently
crystal never copies data, instead streaming it, reducing energy consumption and the need for storage space. It’s like having a compact data center running on low computing power in your pocket.