AI Biweekly: Green Tech - What if computers needed to be refueled?

Dear Artificial Intelligence Enthusiasts,

Recently, the "Green Software Foundation", which brings together big names in the world of Tech like Microsoft, Accenture, BCG GAMMA, and many start-ups, launched a series of events to raise awareness of the growth of tech's carbon emissions. This is a fundamental step, given that globally, 5% of CO2 is emitted by computers – a figure already higher than emissions from the aviation sector! – and that this figure is growing exponentially.

Cars have considerably reduced their emissions in recent years thanks to innovations and new habits: less polluting combustion engines, the revolution of electric vehicles, more alternatives to owning a private car, and, finally, new generations of eco-drivers with their eyes fixed on their consumption data.

In comparison, the world of technology is a mass ecocide... More and more data, computing power, screen space, higher resolution, etc. The immediate consequence: Emissions related to digital technologies are exploding! Why? Because computers don't need to be refueled at the gas station.

Who even knows the cost of their electricity consumption caused by digital tools? Who really is concerned about the greedy and frugal equipment for their phone, computer, or tablet, like car buyers usually are? Who slows down their use of digital technology for environmental or financial reasons?

Almost no-one, because it is so difficult to estimate the electricity consumption and CO2 emissions associated with digital uses. The pooling of infrastructure (telecommunications networks, servers, etc.) complicates the allocation of emissions to users and uses. In addition, emissions related to electricity consumption depend on location and time.

Yet tech players are starting to move. We now have an index of "Software Carbon Intensity" to encourage developers to work as accurately as possible. A "pre-trained" NLP software has just been developed, with a performance of 96% compared to the best benchmarks, and a computing power consumption eight times lower... New tools are helping data scientists measure their emissions to encourage them to "eco-train" their algorithms. "Carbon aware applications" are multiplying to give real-time indications to users in order to reduce their consumption. In terms of infrastructure, hosting providers are encouraged to make the emissions associated with each use more readable as well as to increase competition for access to carbon-free electricity.

The priority must be to raise awareness among users, both professionals and individuals.

We are at the very beginning: We all know that we have to turn off the computers at night – who would run the engine of his car run in the garage? However, our digital behavior is often equivalent to picking up our children from school in a big truck...?The mismanagement must stop.

Until next time,

Sylvain Duranton?

++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++

More articles on the topic:

How can we reduce the carbon footprint of global computing??- MIT News

The voracious appetite for energy from the world’s computers and communications technology presents a clear threat for the globe’s warming climate. At a two-day workshop, hosted by MIT’s Climate and Sustainability Consortium (MCSC), MIT-IBM Watson AI Lab, and the Schwarzman College of Computing, participants worked on solutions to help reduce the carbon footprint of IT infrastructure.

AI Tool Aims to Help CIOs Monitor Carbon Emissions from Cloud Operations - Wall Street Journal

Cloud-management platform Cirrus Nexus launched an AI tool to help companies measure and reduce carbon emissions from their cloud operations, an area of increased focus as companies face pressure from shareholders and regulators to meet sustainability goals.

The Imperative for Sustainable AI Systems – The Gradient

AI systems are compute-intensive: the AI lifecycle often requires long-running training jobs, hyperparameter searches, inference jobs, and other costly computations. They also require massive amounts of data that might be moved over the wire, and require specialized hardware to operate effectively, especially large-scale AI systems. All of these activities require electricity — which has a carbon cost. There are also carbon emissions in ancillary needs like hardware and datacenter cooling.

The current state of affairs and a roadmap for effective carbon-accounting tooling in AI – devblogs.microsoft.com

Digital services consume a lot of energy,?and it goes without saying that in a world with accelerating climate change, we must be conscious in all parts of life with our carbon footprints. In the case of the software that we write, specifically, the AI systems we build, these considerations become even more important because of the?large upfront computational resources that training some large AI models consume, and the subsequent carbon emissions resulting from it. Thus, effective carbon accounting for artificial intelligence systems is critical!

Why public clouds lead with renewables - InfoWorld

It’s no surprise that the top tech companies are also the top cloud companies. It’s also no surprise that these top tech companies are making big investments in green energy. Amazon, Microsoft, and Meta are now the three largest buyers of solar and wind power, according to new data from BloombergNEF, and they easily outdistance all other corporate buyers.

Carbon and the cloud: Why data may be part of the climate change problem - Venturebeat

In the “Exponential Climate Action Roadmap” laid out by the?World Economic Forum, digital technologies could help reduce global carbon emissions by up to 15% – one-third of the 50% reduction required by 2030. On the surface, this sounds like a lofty goal. But, reducing your environmental footprint first requires understanding how your data storage is currently managed, both from an IT perspective and across other departments within your organization.

How to Fight Climate Change as a Software Engineer - InfoQ

Huge amounts of energy are wasted every day by software blocking space and consuming energy at data centers without being used most of the time. We need to consequently scale software down to zero and remove unused deployments from data centers. It is worth taking a look at the actual resource consumption of software; efforts to reduce this resource consumption pay off in terms of lower energy and hardware consumption. The impact looks small initially, but scaling effects turn it into significant numbers.