Image: © LuckyStep/Stock.adobe.com
Kyndryl’s Tim Martin believes that generative AI could be used to reduce its own environmental impact – but only if deployed correctly.
Generative AI (genAI) has the capability to change the world as we know it. In many ways, it’s the revolutionary technology of our time – capable of transforming industrial and business sectors at an unprecedented speed and scale.
But in the race to adopt and deploy genAI across their workflows, few organisations have paused to rigorously assess the environmental impact of this new technology. Besides the widely reported risks related to the use of AI, the massive amounts of energy needed to power these systems poses a problem.
Satisfying the power requirements to drive the revolutionary technology of our time, directly contradicts our efforts to tackle the climate crisis – one of humanity’s greatest collective challenges.
Recent estimates indicate that current AI technologies could consume energy on the scale of nations. For example, experts estimate that the infrastructure required to support some genAI systems could use upwards of 29.3 terawatt hours (TWh) annually – equivalent to Ireland’s annual energy consumption. We need to act now if we’re to continue to drive progress in genAI without creating further environmental damage.
Each data centre requires a large amount of energy for computing, data storage and cooling. To reduce greenhouse gas (GHG) emissions, optimising energy management in these systems is one of the ways to contribute to this reduction.
In addition to technical changes, organisations should be considering strategic adjustments. Improving energy efficiency is not enough; we must consciously integrate sustainability into the planning and deployment of these technologies.
But what if we could use genAI itself to manage and reduce its environmental impact? We know the technology has the capability to optimise operational processes, automate management systems and provide detailed visibility into workflows. So why not apply these capabilities to genAI programmes?
By leveraging genAI to track broader AI functions, organisations will be able to quantify their carbon footprint and identify opportunities to significantly reduce it. As the technology evolves, genAI may be able to optimise resources to neutralise its net impact on the environment – perhaps even balancing the CO2 it generates.
On a technical level, organisations need to have smarter systems architectures in place to make the most efficient use of genAI. Optimising code, using specialised hardware and migrating workloads to the cloud are all strategies that can help reduce energy consumption significantly.
And from a business perspective, it’s critical for organisations to take a holistic view of their technology operations to prioritise solutions that not only increase efficiency, but also align with sustainability targets.
Migrating workloads to clouds managed by specialised providers may offer tangible opportunities to reduce energy consumption. Modern data centres optimise energy use through advanced techniques such as server virtualisation and automation.
Innovations in cooling these data centres are also helping to reduce energy consumption. Features such as mechanical chillers, passive systems and direct liquid cooling all contribute to these efforts.
Leaders must join with developers and scientists to take responsibility for AI and its outputs. It is a collective responsibility for us all to ensure that this powerful technology is deployed ethically, responsibly, efficiently and in ways that aren’t counterproductive to environmental progress.
Used responsibly and strategically, genAI can help us to build a more sustainable future. But to achieve this, companies must look beyond short-term technology fixes and commit to a longer-term, considered vision that doesn’t prioritise innovation at the expense of the environment.
By Tim Martin
Tim Martin is the director of environmental management at Kyndryl.
Don’t miss out on the knowledge you need to succeed. Sign up for the Daily Brief, Silicon Republic’s digest of need-to-know sci-tech news.
