Despite being a relatively new phenomenon, artificial intelligence (AI) is rapidly appearing in modern-day life as technology replicating human actions. That may take the form of chatbots like ChatGPT, image generators, voice recognition like Siri or Alexa, or even driverless cars. Because using AI is instantaneous with no evident emissions associated with its use, the high volume of water and energy used in data centres remain invisible to many. Sustainability is about protecting the future, and since AI seems to be here to stay, let’s examine how sustainability and AI are intertwined.
The training of AI allows it to memorise vast quantities of data, yet it also relies on an enormous amount of computing power, creating an environmental impact. At the same time, AI systems are expected to help prevent climate change, with research predicting the application of AI levers could reduce worldwide greenhouse gas emissions by 4% in 2030, equivalent to the emissions of Australia, Canada, and Japan combined.
AI’s role in preventing climate change and protecting biodiversity
Let’s start with the positives. A 2020 study assessing the impact of AI on the UN Sustainable Development Goals found that AI could enable 93% of environmental targets. This included using AI in creating low-carbon cities and integrating renewable energy into clean smart grids, reducing resource consumption and energy wastage. Another potential for AI is tracking environmental changes — such as desertification, marine and air pollution, weather changes — through satellite imagery, and preventing forest fires. Artificial intelligence has been identified as a key technology for conservation, reducing the manual labour required to collect data on protected species and stopping poachers, helping to tackle the biodiversity crisis.
While many are preoccupied about AI stealing jobs, if harnessed for environmental applications AI could create 38.2 million new jobs globally, contributing up to $5.2 trillion USD to the global economy in 2030 (a 4.4% increase to business as usual).
So-called ‘Green AI’ has vast potential for use in promoting circular economy practices, keeping products in use and tracking the lifespan of materials such as recycled plastic. It could also help decarbonise supply chains by offering end-to-end transparency, something that excites us at Grain as it would help businesses tackle scope 3 emissions.
The environmental impact and ethics of AI
Now for the not-so-good. A recent study claimed the creation and training of GPT-3 by OpenAI was predicted to have a carbon footprint equivalent to driving a car from Earth to the moon and back, as well as using 3.5 million litres of water cooling data centres. As datasets grow and algorithms become more complex, this energy and water usage is only expected to increase, unless more efficient data processing techniques are discovered. It is claimed that many AI creators are not currently reporting transparently on the energy usage of their systems, making the true impact hard to quantify.
When considering the sustainability of artificial intelligence, it is important not to bypass the social impact and ethical dilemmas this technology brings up. There are fears that AI could exacerbate social inequality. Due to a lack of accessibility, the data collected may come from narrow cultural perspectives and replicate human prejudices such as racism and sexism. As AI decisions lack human intuition and oversight, the ethics and wider implications of its actions may have negative consequences, along with personal data breaches. Human rights issues have also been highlighted in the creation and maintenance of AI devices — for example, programmers may work in sweatshop-like conditions.
It is evident that AI has many positive uses that could assist in tackling climate change and boosting sustainability in many industries, yet the negative impacts of its use should not be overlooked. If we want to use AI to save the planet, we also need to consider its impact. Artificial intelligence will be a positive tool if we can ensure the transition to ‘green’ and ‘ethical’ systems, and focus on reducing the energy intensity of its use through technological developments and renewable energy. Google’s DeepMind division has recently trained a system that has taught itself to conserve energy by 40%, showing that with responsible application, this new and growing power has immense potential.