Data Pollution
The Hidden Environmental Costs of Data and AI
An email instead of a letter, online shopping instead of driving to a mall, a video conference instead of an in-person meeting. Are these activities as green as we think, or do they hide an environmental footprint?
Data has become an integral part of our daily lives, helping us to make better decisions and to improve our quality of life. However, the increasing amount of data that we generate and process requires an enormous volume of energy, which has negative social, economic, and environmental impacts.
Numbers
80-130 million gallons
Amount of water consumed per year by data center
138.000 tons
CO2 from top websites cookies per year
65.4 megatonnes
CO2 from the Bitcoin network per year
26 kg
O2 from search queries per year & person (5 billion internet users)*
8-15%
ad related activities in top 350 sites (1 ad impression: 1 gr emissions)*
5%
amount of storage data that is used, the rest is digital waste*
23 kg
CO2 from ChatGPT’s estimated daily carbon footprint, equivalent to a car consuming 10 liters of gasoline*
*Non-Eticas data
The environmental impact of data centers
Data centers store all the digital information in massive facilities that need huge amounts of electricity to power them, and water to keep them cool (data centers with 15 MW of IT capacity can consume between 80 and 130 million gallons per year). They can have both positive impacts, like job creation and investment, but they also amount to more than 2% of the world's electricity consumption, and generate the same volume of carbon emissions as the global airline industry. To that, we need to add the waste they generate. There are some efforts being made to reduce the environmental impact of data centers, but overall, they need to become more sustainable in order to mitigate their negative impact on the environment.
Lowering polluting data
There are mitigation measures to lower data pollution. One of them is data minimization, which consists of collecting only the minimum amount of personal data necessary to achieve a specific purpose. Apart from reducing pollution by reducing the amount of information stored, minimizing data can also improve the quality of data processing outcomes. Other mitigation measures include green Machine Learning methods like AI modular blocks to simplify training processes, reducing the amount of time and energy required; spam filtering to reduce the quantity of emails received saving up to 135 TWh of electricity per year; and PoS replacing PoW for Crypto pollution. Regarding data centers, using renewable energy sources and implementing energy-efficient technologies are some of the solutions that are beginning to be implemented.
Recommendations
Raising awareness regarding the environmental impact of data processes is key to pressure governments and companies to tackle data pollution.
To fairly address data pollution, global experts must come to an agreement regarding measuring standards.
The implementation of mitigation measures like data minimization, green ML methods, spam filters, or data centers sustainability is a good start.
Advocating for the inclusion of the ICT industry in the EU Emissions Trading System would create market-based mechanisms to reduce emissions, and help analyzing the impact of future technology implementation.
Purchasing carbon credits or tax breaks could be incentives to improve companies’ reputation and show their commitment to sustainability.
Considering data centers as physical property to develop innovative tax regimes (e.g. compensation schemes based on distribution of dividends)
The environmental cost of the vast amount of data processed everyday needs to be acknowledged. When we agree on how to measure data pollution, we will be able to find solutions to make technology a true green option that doesn’t contribute to climate change, but helps us fight against it.
Latest
Quick Quiz: When Generative AI is asked to identify gender based on professions does it:
Read More
The Hidden Cost of Our Digital Lives: Understanding Data Pollution Our digital world comes with
Lawmaker or Lawbreaker? How FaceNet Got It Wrong Would you trust
Tech has helped throughout the years mitigate climate change, from solar and wind energy to
Sweden hosted the fourth ministerial meeting of the EU-US Trade and Technology Council (TTC) in
Period-tracking apps are part of what is known as Femtech, a growing industry which has
Tech regulation has struggled with enforcement for a long time. Drawing on our experience with
Should faces be used to make decisions on insurance premiums? This adversarial audit report looks
A tool designed to assess inmates' risk of recidivism should be robust and reliable. This