Recently, artificial intelligence (AI) has been presented as an issue in urgent international efforts to combat climate change. As AI plays a much bigger role in our livesit’s going to require enormous amounts of computing power and data storage.
Hence the CO2 footprint of AI is anticipated to be expanded as a consequence of its high energy consumption and the carbon emissions related to the production of its hardware.
However, the reality is more nuanced, as artificial intelligence may be an issue solver, making a major contribution Contribution to combating climate change.
For example, AI could help to be more accurate Predictions of maximum weather events like hurricanes or the speed at which we will expect the world's polar ice and glaciers to melt. It could also help us higher manage our energy infrastructure, equivalent to power grids.
Since 2012, the biggest AI training processes have been consuming increasingly computing power. In fact, the speed has doubled on average every 3.4 months.
Data centers and transmission networks contribute to this over 1% of world energy consumption and 0.6% of world carbon emissions. A single query ChatGPT (OpenAI's advanced chatbot) can generate rather a lot more carbon than a traditional Google search.
Jon Rehg/Shutterstock
Use AI for good
Despite the possible negative consequences, there may be reason for optimism. One way AI could assistance is: improve our understanding the underlying science of climate change.
There are some ways AI could contribute on this regard, but one could possibly be one by improving climate models. These are computer-based simulations of how the Earth's climate works and the way it responds or will respond in the long run to increased concentrations of greenhouse gases.
AI techniques equivalent to machine learning could possibly be used to simulate elements of climate models which can be currently difficult to breed, equivalent to the formation of raindrops or clouds.
In this fashion, artificial intelligence couldn’t only improve the forecasts from climate models that function the premise for political decisions, but additionally reduce the computing power required for such tasks. This, in turn, could help reduce the carbon footprint of running these climate models on supercomputers.
In a recent TED talkSims Witherspoon, head of climate and sustainability at Google DeepMind, said AI can assist us optimize and manage existing systems and infrastructure equivalent to power grids. Electrical networks should be actively managed to keep up a stable balance between supply and demand.
Transitioning from fossil fuels to renewable energy is critical to achieving net zero targets. However, while fossil fuel power plants are relatively reliable, wind and solar energy is unpredictable as a consequence of weather. That is where AI can potentially intervene.
Witherspoon's team at DeepMind trained a neural network (an AI system inspired by the human brain) using data from historical weather patterns and data about electricity production from wind turbines. The resulting technology performed 20% higher than existing wind power generation forecasting systems. This could possibly be utilized by operators to higher plan for supply shortages and fill the gaps with energy from other renewable sources.
The right guidelines
Despite the tangible advantages, AI also requires the appropriate government policies to comprehend its potential. The interplay between the advantages of AI and its environmental costs is incredibly complex. To make sure that AI has an overall positive impact, all stakeholders, including governments and technology firms developing AI systems, must ensure transparency about their environmental costs.
Only through transparency and data sharing can we make informed and strategic decisions concerning the use of AI to amplify its positive impacts and find solutions that reduce the technology's harmful impact on the environment.
As things stand today, AI shouldn’t be a very environmentally friendly technology and its development is dear. However, climate change is our biggest challenge and AI could possibly be a precious ally as we discover solutions that offset its disadvantages.
