The Environmental Impact of AI

In our era, two significant paradigm shifts have emerged - combating climate change and digital transformation.

Since the dawn of the web in the late 1980s, we have witnessed a technology revolution that has changed the way we think and work. Where Yahoo! Search was among the most popular search engines in the 1990s, Google Search took the mantle in the 2000s and continues to do so even today. Today, Google Search has a global market share of more than 90%. But the evolution of technology is not limited. It grows by leaps and bounds on a regular basis. 2022 saw ChatGPT, the modern-day chatbot driven by Artificial Intelligence, get introduced into the market and it has become one of the most talked-about tools of present.

AI has been incorporated in many functions and its usage has been rising considerably over the last few years. This rise is also showing promising signs of being able to support climate action. However, it is crucial to understand its net potential and be vigilant about its rapid scaling. Google, in a report released in November 2023 in collaboration with Boston Consulting Group, opined that AI has the potential to mitigate 5-10% of Global GHG emissions by 2030. But with AI being integrated into search engines, more computing power is needed to train and run models. Experts say this could increase the computing power needed as well as the energy used by up to 5 times per search.

Potential of AI in climate action

AI offers several opportunities to combat climate change and promote sustainability. With its ability to optimise processes and reduce energy consumption, AI has the potential to accelerate climate action. Here are a few examples:
 

• Energy forecasting: National Grid ESO (UK) has used AI to double the accuracy of its forecasts of electricity demand, facilitating a better integration of renewables.
   

• Simulating urban climates: The InFraRed project at the Austrian Institute of Technology uses AI to model the effect of urban design on the microclimate in cities, allowing simulations to run in seconds instead of hours. The impact of such methods can be boosted via standards for integration of AI tools with design platforms.
   

• Advancing battery design: The US-based startup, Aionics, leverages AI to design batteries with improved efficiency and 10x speed increase, addressing a critical bottleneck to the wider adoption of electric vehicles and renewable energy.
   

• Locust outbreak prediction: Kuzi, a tool by the Kenyan company Selina Wamucii, uses AI to predict locust outbreaks, a devastating threat to farmers that has intensified with climate change.

While AI shows promise in climate action, there are also many concerns about its environmental impact that need to be addressed. These include:

• Increased energy consumption: AI systems demand substantial computational power, resulting in increased energy consumption. Data centres run 24x7 and most of them derive energy from fossil fuels.

  
  
  

• Increased water usage and cooling: ChatGPT is estimated to require 500ml of water for every five to 50 questions answered. Microsoft's data centre water usage increased by 34% from 2021 to 2022, consuming over 1.7 billion gallons, equivalent to filling more than 2,500 Olympic-sized swimming pools.

  
  
  

• Carbon footprint: The manufacturing and disposal of AI hardware contribute to electronic waste, while sourcing materials like rare earth metals can harm ecosystems. Additionally, the carbon footprint associated with the life cycle impacts of AI software and hardware must also be considered.

To mitigate the environmental impact of AI, responsible practices need to be incorporated. These include:

• Developing energy-efficient AI hardware and algorithms

  
  
  

• Transitioning data centres to renewable energy sources

  
  
  

• Companies can prioritise sustainable design and procurement of AI systems to minimise their environmental impact throughout their lifecycle

As AI continues to grow rapidly, it is imperative to assess its impact on the environment in a comprehensive manner. Governments, international frameworks, policy makers could consider incorporating climate impact considerations into AI regulation and mandating responsible AI principles into design structures. They can also promote research into developing energy-efficient hardware and algorithms for AI systems can help reduce their carbon footprint. Sectoral government data portals for climate-relevant sectors in lower income countries can enable equitable access to data and support easy access to the data required for high value climate problems.

We are in a climate emergency. To mitigate this, governments, public and private sectors need to come together and bring change. AI, by its very being, presents both opportunities and challenges when it comes to its environmental impact. While it can accelerate climate action by optimising processes and efficiency, the increased energy consumption, water usage, and electronic waste associated with AI cannot be ignored. What the tool needs is regulation and sustainable approached to AI development. Through this, we can harness its potential while minimising its environmental consequences to drive positive climate outcomes.