Algorithms under the Bonnet: Reaching the Goal with Less Energy
Doing the same with less energy: Over the next three years, the TANGO project will work on a cross-sectoral data platform. One goal is the development of particularly energy-efficient algorithms. Why energy-saving AI belongs under every bonnet. And how the eco – Association of the Internet Industry supports the project and its goals.
Steering, accelerating and braking autonomously – sensors, cameras and IT systems in smart cars make exactly that possible. What is technologically necessary to drive safely and comfortably has an impact on the climate; and significantly so, as a study by the Massachusetts Institute of Technology (MIT) shows. According to the study, the computers of autonomous cars will soon be responsible for around 0.3 per cent of all global carbon dioxide emissions annually. This is as much as all the data centres in the world currently emit in the same period. The MIT forecast assumes 1 billion cars driving autonomously for 60 minutes a day, each consuming 840 watts of computing power. There are currently only around 55 million cars of this type on the roads of the world. But nevertheless, a clear trend is emerging. The automotive industry is already interested in solutions such as those now being researched by MIT. On the one hand, the hardware can be tailored to be specifically suitable for the computing tasks in the vehicle. And secondly, the software can be optimised so that the algorithms under the bonnet consume less energy.
The TANGO project: A platform for ecologically sustainable data management
Achieving the same result with less energy; intelligent software with a smaller ecological footprint is one goal among many others pursued by the TANGO project launched in September 2022. Under the title “Digital Technologies ActiNg as a Gatekeeper to information and data flOws”, 34 partners from 13 countries are working together under the leadership of ATOS Spain. By 2025, the consortium will realise a platform based on a citizen-centred, secure and trustworthy concept that promotes transparent, secure, fair, responsible and also environmentally sustainable data management.
Speed up calculations, improve forecasts and save energy
TANGO develops particularly energy-saving neural networks. The programmes, which are modelled on the human brain, are necessary for Deep Learning and thus crucial for Artificial Intelligence (AI). Because AIs can be trained and further developed via neural networks. Such autonomous learning machine applications (AutoML) are able to accelerate calculations and improve forecasts independently. Typically, this makes it in demand wherever large amounts of data have to be processed. And typically, the computationally intensive applications consume a lot of energy in the process – whether on TANGO’s data platform or in the autonomous car. “TANGO aims to offer an AutoML solution designed to generate energy-efficient deep neural networks that also provide appropriate performance in terms of accuracy, considering that the training or inferencing of the models should be done in distributed environments and include edge computing devices,” the project website states.
Study from Portugal: Programming languages as power guzzlers
Software in the cloud: Opportunities for the environment and climate
Software as a climate factor – a study by Accenture highlights the connection. To do so, the consulting firm examined the life cycle assessment of the public cloud in 2020. The results make it clear: If companies worldwide migrated their IT to provider data centres, 5.9 per cent of all emissions would be eliminated that would otherwise be caused by local operation. The move could save around 60 million tonnes of greenhouse gas annually, equivalent to the amount emitted by 22 million cars over the same period. Around 12 million tonnes of potential savings are accounted for by software when applications are programmed to take advantage of cloud-native opportunities.
Cool more efficiently, utilise resources better and save energy – these advantages of shared infrastructures show why cloud data centres are superior to any dedicated server room. And opportunities that the TANGO project also addresses: “With climate change being the defining issue of our time, operators are now quickly realising that there is an urgent need to increase the energy efficiency of data centres and reduce their environmental impact.”
EU spends one in three Euros on climate targets
eco – Association of the Internet Industry is also working on ways to combine the digital with the green. And not only as a consortium partner in the TANGO project, but also with its own initiatives: For example, the Alliance for the Strengthening of Digital Infrastructures in Germany has been formed under the umbrella of the Internet association. According to figures published by the alliance of cloud providers, Internet service providers, software manufacturers and other companies from operating industries, the CO2 emissions from European data centres have been declining since 2015. The trend is driven by advances such as those TANGO is now striving for.
TANGO’s work nestles strategically within Gaia-X’s decentralised and distributed data infrastructure. And TANGO is part of Horizon Europe, the European Union’s largest research and innovation programme to date, worth €95.5 billion. Every third Euro is to flow into technologies that help the confederation to achieve its climate goals. Digital solutions as a building block for a healthier and more sustainable world – after all, this is not the only thing the Massachusetts Institute of Technology is convinced of. Because if autonomous cars are on the road with more energy-efficient algorithms, this is not only more ecological, but also more economical: If on-board computers consume less electricity, the range of electrically powered vehicles increases.