One of the last things we might think about is its carbon footprint when we think about AI. Yet, this footprint is enormous.
By one estimate, quoted in a report on AI by accountancy body Chartered Accountants ANZ, the computational power needed to train the largest AI models has grown exponentially, doubling every 3.4 months.
Another study showed that the carbon emissions for training a single natural language processing model were equivalent to 125 round-trip flights between New York and Beijing.
And more complex algorithms such as GPT-3, used for language and text analysis, use even more energy.
When the algorithms have finished doing their thing, the energy used by computing systems eventually dissipates as heat. So, large data centers running AI algorithms need to have significant energy for cooling systems.
Data centers, therefore, end up consuming energy by the nature of their activities, and conventional production of that energy results in the emission of greenhouse gases.
In addition, toxic materials or rare earth metals may be needed for cooling or other support processes.
Turning a deaf ear
With all this in mind, the Green IT movement has gained some momentum in recent years as technology professionals understand they can also contribute to reducing emissions.
But as a recent report from Gartner shows, the message is only getting through slowly.
Although around 80% of large enterprises currently produce regular sustainability reports, surveys show that only 26% actively pursue green practices across IT.
Part of the problem lies in the challenges of measuring the material footprint of technology, baselining current progress, and establishing clear paths towards goals.
Just as in any supply chain, aligning in-house sustainability standards with the ecosystems of vendors can also be difficult to quantify and communicate.
IT is increasingly outsourced or consumed as a service from the cloud, which creates issues in measuring the material footprint of much of the technology an organization uses.
The commitment of vendors to sustainability is a “black box” to sourcing, according to Gartner.
CTOs are your sustainability kings
There is some hope. And the message is permeating as many chief technology officers step up and become sustainability leaders.
According to the 2020 Gartner Sustainability Survey, 42% of the surveyed IT and business leaders whose organizations have a sustainability strategy say their CIO or CTO plays a critical role as either a sustainable business enabler or technology innovator.
Furthermore, 63% of the organizations already invest in IT and digital solutions as part of their sustainability strategy.
“For CTOs, sustainability has the potential to create greater operational resiliency and financial performance while acting as an ambitious technology innovation opportunity focused on exploiting new avenues of growth,” the Gartner report says.
“That being said, sustainability does not guarantee these business or technology performance benefits. In certain situations, it may cost more money to support a sustainable goal and may impair operational resiliency in scenarios where the organization uses energy-efficient tools that are less productive or more technically challenging,” the report adds.
In other words, sustainability goals can be another compromise or trade-off in the IT mix, adding another complexity level.
Need to tie green goals to digital leadership
According to Gartner, the way forward is for organizations to implement strategies that combine environmental and social benefits with the proactive identification of areas to increase digital leadership.
The Gartner report suggests a five-level maturity model as organizations move towards sustainability. It begins with awareness, moves to the definition, and then measurement, which leads to management and optimization.
There’s no doubt that the momentum is growing. By 2025, 75% of CTOs will be tasked with contributing to enterprise sustainability goals, and 25% will have metrics tied to sustainable progress, specifically around carbon neutrality.
Gartner’s advice is to select no more than four key goals and limit the number of projects to achieve those goals to around 10 or 12.
The three key outcomes of greener IT are improved operational efficiency, enhanced brand reputation, and innovation around new products.
But at the same time, CTOs cannot conceivably tackle all the potential areas of sustainability where they might have influence.
The danger lies in tackling too many goals and spreading the effort too thinly, which results in the sin of greenwashing.
Instead, CTOs should focus on what they can control and go hard in several definable areas. This will result in a smaller carbon footprint for the organization and some lasting operational benefits as well.
Lachlan Colquhoun is the Australia and New Zealand correspondent for CDOTrends and DigitalWorkforceTrends, and the editor of NextGen Connectivity. His fascination is with how businesses are reinventing themselves through digital technology and collaborate with others to become completely new organizations. You can reach him at [email protected].
Image credit: iStockphoto/Vasyl Dolmatov