Artificial intelligence (AI) is one of the most consequential technologies today. It is dramatically changing the world, altering business models and ushering in transformational development in education, healthcare, communication and agriculture, to name a few. It has introduced efficiencies to our workplaces, improved industrial productivity, optimised resource use and reduced waste. AI’s rapid rise is attributed to availability of data and a phenomenal growth in computational power.

AI has also furnished new tools to address environmental issues and climate change. From spotting smog in the sky and detecting deforestation on the land to minding microplastics in the ocean, AI’s potential is tremendous. AI-based early warning systems predict cyclones, or rains and minimise the extent of damage. Forecasts about droughts and heatwaves provide timely information to vulnerable communities. AI-generated techniques in smart agriculture help farmers cope with water stress, prevent waste, optimise crop yield and ensure food security.

But AI can also harm our fragile planet. AI is power hungry. Data collection, storage and application require an elaborate system of energy-intense data centres. A rise in AI deployment means a rise in the number of data centres with a commensurate demand for power. A recent Internat­ional Energy Agency meeting noted that the “electricity demand for data centres and digital services would likely rise substantially in the next few years, despite continued improvements in the efficiency of hardware and software”. To operate under controlled temperatures, data centres need air-conditioning consuming more energy. The United Nations Environment Programme’s (UNEP) recent ‘Issues Note’ on AI points out that the number of data centres worldwide has grown from 500,000 in 2012 to over eight million, with energy consumption doubling every four years.

With the current global energy mix still dominated by dirty fuels, this trend can translate into a spike in greenhouse gas (GHG) emissions. In the process, our planet will be the unintended victim. Until a just energy transition is expedited, AI’s carbon footprint is set to increase.

AI’s appetite for power is matched by its thirst for water. To cool the heat generated by energy-intensive servers, large quantities of water are needed. Global demand for water for AI needs is estimated to reach four to six billion cubic metres by 2027.

AI’s potential negative impact on the environment goes further. Critical minerals are an essential part of hardware for digital technologies. Found mostly in high-water-stress areas of developing countries, such mining, without efficient mitigation measures, puts more pressure on natural resources, damaging biodiversity, polluting air and spewing GHG. Waste and e-waste produced during the life cycle of digital technologies is another emerging problem.

Fortunately, these problems can be solved through a strategy that incorporates AI’s pros and cons. The Summit of the Future last September acknowledged AI’s promise to accelerate progress across the SDGs and address its possible negative impacts on the environment and other areas. But how can this promise be realised urgently? The missing link between AI’s potential and a strategy to extract its full advantage is a regulatory regime.

International discourse on the issue has generally revolved around matters of privacy and disinformation, etc, but environmental costs are now part of it. Governments and tech companies—leaders in deploying AI at scale—concede that environmental impacts will need to be included in any discussion and tackled urgently. An intern­a­­­tional governance system is needed for the gamut of subje­cts in the AI realm.

The first steps were taken in this direction at the UN last year. A global dialogue on AI governance has been initiated and an independent international scientific panel established to produce evidence-based impact, risk and opportunity assessments. To aid these efforts, UNEP’s ‘Issues Note’, contains several commendations, including establishing standardised methods and metrics for measuring AI’s environmental impact and developing a mechanism for mandatory reporting and disclosure of AI’s direct environmental impacts by tech companies, focusing on consumption of energy, water, mineral resources, carbon emissions and e-waste.

The environment became a victim of the consequential technology of the previous age—the Industrial Revolution, that is wreaking havoc to this day. If it becomes a victim of the runaway impacts of the consequential technology of this age, it can knock down the planet. For harnessing AI’s full potential, it is imperative to assess and manage its associated risks and mould it into a shield for the planet.-

-Dawn (Pakistan)/ ANN