Some of the issues and problems which need to be considered when deciding if, and how, to use AI.
Reflection points
- What will AI mean for students?
- What will they need to know about AI to be informed users of these new technologies?
- How is it possible to balance the ethical and environmental issues and problems with the societal and economic implications of AI?
Sustainability
Artificial intelligence will change the way the world processes and uses information. It will make it easier and quicker to generate complex and seemingly coherent written, visual and musical responses to enquiries. This will transform the way the world does business, manages people’s health and makes new discoveries. It will have an impact on many aspects of peoples lives. However, using AI is not without consequences.
The energy demand placed on the data centres which process AI are enormous and growing exponentially. The World Economic Forum estimates that the computational power for sustaining AI’s use is doubling every 100 days. It is estimated that, currently the data centres processing AI are consuming between 1-2% of the world’s power. It is believed that this will rise to between 3-4% by 2030.
The International Energy Agency suggests that a ChatGPT query requires 10 times as much energy to process as a Google search at 2.9 Wh (watt hours) compared to 0.3Wh. The number of people using AI tools was estimated to be over 250 million in 2023 and is forecast to grow to 700 million by 2030. In October 2024 ChatGPT set a new record of 3.7 billion global visits, up 17.2% month-over-month. and up 115.9% year-over-year.
There are thousands of data centres around the world and their demand for power is growing markedly. These centres, not only require huge amounts of electricity, they also require cooling. This consequently places a growing demand on water supplies. For instance, an average of 25 million litres of water is required annually for each single megawatt (MW) of installed processing power. The data centres in Amsterdam alone have an installed capacity of 950 MW. A recent study suggested that AI related infrastructure worldwide may soon consume as much water as Denmark.
It is also necessary to factor in to the escalating Costs of AI, the raw materials required and moreover the amount of e-waste such as mercury and lead that is generated. So AI comes at a considerable cost to the planet.
Issues and problems, signs of change
This is a new industry, however, and steps are being taken to mitigate the issues and problems, such as, the drain on resources and improve sustainability. Apple’s own data centres now run on 100% renewable energy. In 2023, Apple’s servers saved 12 million gallons of water, due to the fact that five of Apple’s owned and operated data centres were certified to the AWS standard. The remaining will be certified in 2025. Microsoft’s Azure meanwhile aims to use 100% renewable energy by 2025 and is focussed on being water positive by 2030.
In the European Union, data centres consume almost 3% of the EU’s energy. However, steps are underway to reduce and regulate energy and water use. Operators of the larger data centres will now measure their energy efficiency, cooling effectiveness and the proportion of renewable energies that they use. In the UK, data centres are classed as Critical National Infrastructure , therefore, the government can ensure they minimise the impact data centres have on water supplies and electricity.
In January 2025 the UK government announced a new blueprint to use AI to drive an action plan for change and to help turbocharge growth and boost living standards. Click here to see the 50 recomendations of the Action Plan.
Decisions about if, when and how to use AI need to be taken with an awareness of cost and an eye to the future.
Sources include: David Crookes, How Green is your AI, Mac Format issue 411 December 2024
further reading