A Look at Digital Tech's Environmental Ledger
Two years ago an analysis from the OECD shared:
“digital transformation is increasingly recognised as a means to help unlock the benefits of more inclusive and sustainable growth…in the environmental context, it can contribute to decoupling economic activity from natural resource use and their environmental impacts.”
This idea aligns with the "Tech or AI for Good" narrative, championing the belief that emerging digital technologies can help us achieve UN Sustainable Development Goals – advancing health, education, climate action.
What’s lost in this narrative is the upside is intertwined with downside, 1) the real costs of generating emissions by manufacturing devices, storing large amounts of data, training the models and 2) further enabling bad forces (think disinformation, war, mining extraction, algorithmic consumerism). A few points to build this picture:
We now use the same energy demands of Japan to power datacentres and soon to be 2X
A recent study in nature journal ran a lifecycle assessment finding that current digital content consumption requirea on average 41% of the per capita carbon budget consistent with a high likelihood (67%) of limiting global warming to 1.5 °C above the pre-industrial levels
Microsoft's reported emissions have increased by 30%, despite a self-imposed target of reducing emissions by 30%
This tension between good and bad is not a full paradox, there’s a midpoint, an opportunity to moderate and avoid harm. We can use data to decarbonise worthy things by providing decisioning support to help navigate complexity & uncertainty in sustainability transitions , dial down videos on midjourney and ensure the main drivers of emissions all run on renewables. So how should we assess the value, where should we deploy it, how should it be governed and what should we expect them to do?
‘The beginning of wisdom is the definition of terms’ Socrates
Firstly lets merge the definitions of data products, software and AI together to talk at a high level about data and information. What we are talking about are technologies that improve our information and communication system [Lyytnien].
With each new wave we are adding a more sophisticated layer of information and are building new ways to track and measure risk, surface knowledge and understand all the things around us. With the advent of more sophisticated tools the feedback loops are quicker, in real time and can comprehend more complicated problems by applying math to large and unstructured data and give another good crack at the dark arts of predictions.
In these modern information systems, software can be seen as databases with user friendly experiences coded by humans, and artificial intelligence simply means software used by computers to mimic aspects of human intelligence with a focus on patterns and future prediction. This report from Columbia University has a helpful framework diving into detail on the application of AI vs software on climate problems, calling out the main distinctions of the different flavours, software is deterministic vs. statistical, and one is designed for output that is more simple vs. complex problems.
A Ledger of the Good & Bad things to get us to Net Zero
The Arguments for Good Things 😎
#1 Deploying green industry solutions quicker, better, faster
The main promise of this ‘datafication’, or ‘digitisation’, shifts us to the world of ‘Industry 4.0’. With more information, we can decarbonise supply chains, green finance moves us to rewarding lower carbon players with better incentives and starts to turn the tap off for fossil fuel players. There’s been alot of attention on the the digital climate-tech space and some are betting, such as Accenture, that these solutions can reduce global emissions by up to 20 %.
And as net zero becomes policy across the globe, you need alot of excel sheets, measurement and assurance of claims, opening up a significant market for consultancies [economist] and software peaking at $3B investment in 21’, 22’ and $1b thus far in Q1 24’ [PWC climate funding report].
The promise of new players are usually pegged around reimagining ‘workflows’, ‘intelligence’ enabling teams to be more efficient, collaborative and discover whole new pockets of knowledge in the health sciences and climate arena.
This model tends to align with a techno optimist view that we humans are good, the assumption that we always want more information, and that people make better decisions with data rather than human judgement or intuition (Brynjolfsson et al., 2011), ‘what gets managed gets measured’. It is also a reflection that we humans have evolved to really like data, as ‘data’ and ‘measurement’ have become embedded in how we think about creating change. Here's a conceptual model of value being generated by new data enabled tech:
The Arguments for Bad Things 😒
#1 Crazy Data Energy Costs
Emissions from digital products accounted for 3.8% of total greenhouse gas emissions in 2021. In 2040, digital emissions are estimated to increase to 14%, which for comparison would exceed emissions from all of road transport today (11%). Some therefore claim we are on track to therefore exceed the amount of energy that we are building for, and that's a crazy place to be.
Critics talk about how we are far from peak information supply and given the apparent limitless infinities of information and data, that these models just aren’t worth the investment (see Gary Marcus). So a bit like we are trying to do some shredding for summer whilst we eat that basque cheesecake.
#2 Accelerating Misinformation, Fossil Fuel Industries or Consumption
Some advocate caution given the risk of it accelerating bad externalities of our current systems. Examples are enabling disinformation from elections to green claims (fuelled by 90% of VC money is going to media startups), mining companies using it for mineral extraction, making war more opaque and accelerating consumption through SHEIN’s algorithmic business model.
Current regulations appear insufficient to control bad actors, those that are cranking up more fossil fuel activities or more consumption per person. We just need to look back to the late 2000’s for guidance to see a similar situation arise when the economist hailed data as 'the new oil'. Fast-moving tech companies created immense value, but it did not raise all boats. This parallel suggests the need for new approaches to ensure fair distribution of value in the digital age.
Balancing the Scorecard
Research has yet to determine the precise value of digital tools in achieving sustainable development outcomes. Studies at the intersection of digital technologies and sustainability is still in its infancy (Bohnsack et al., 2022) and understanding the impact of data on the integration of green practices on environmental performances are still rare (song et al 2017). There's the debate ongoing over how digital technologies may be an enabler or an additional burden that isn’t really driving sustainable development, and even determined to be paradoxical (Hazas & Nathan 2017), citing the environmental unsustainability of digitalization and hardware production (Bohnsack et al.,; Kuntsman & Rattle,Dieste et al. 2023; Ghobakhloo et al.,2021).
We can also critically look at forecast models for guidance, one scenario is we leverage these technologies to decrease emissions by 20% or on the flipside we crank it up to 14% of total emissions. As with most things in life the outcome generally lands in the messy middle.
In an optimistic scenario, the rapid and efficient implementation of green energy and food systems would be supported by digital technology. A new layer of information would in real time help industry navigate uncertainty and risk whilst decarbonising. This would also enhance green alternatives to showcase their value, enable them to achieve economies of scale to compete effectively with fossil fuel-enabled solutions. And in turn, the finance industry would prioritise green credentials.
However for this scenario to have a chance of playing out, to balance out the downsides of widespread data usage, we require governance structures that address the digital carbon footprint. This would include mandating the use of renewable energy for data centers and hardware manufacturers, optimising energy-intensive cloud processes like video delivery, and establishing a framework that defines "enough" data usage, preventing excessive and destructive practices.
In summary as we are being led down a path, let’s acknowledge both positive and negative aspects and establish some constraints around it.
Thanks for reading
Genevieve
Before you bounce, did you know that there is an olympics for excel? How great.
Bohnsack, R., Bidmon, C. M., & Pinkse, J. (2022). Sustainability in the digital age: Intended and unintended consequences of digital technologies for sustainable development. Business Strategy and the Environment, 31(2), 599–602.
Dieste, M., Orzes, G., Culot, G., Sartor, M., & Nassimbeni, G. (2023). The “dark side” of industry 4.0: How can technology be made more sustainable? International Journal of Operations & Production Management.
Ghobakhloo a, Masood Fathi b c, Mohammad Iranmanesh d, Parisa Maroufkhani e, Manuel E. Morales, 2021. Industry 4.0 ten years on: A bibliometric and systematic review of concepts, sustainability value drivers, and success determinants
Kuntsman, A., & Rattle, I. (2019). Towards a paradigmatic shift in sustainability studies: A systematic review of peer reviewed literature and future agenda setting to consider environmental (un)sustainability of digital communication. Environmental Communication, 13(5), 567–581.
LYYTINEN, 1987. Different Perspectives on Information Systems: Problems and Solutions, Department of Computer Science, University of Jyvisky, Finland
McAfee, A., & Brynjolfsson, E. (2012). Big data: The management revolution. Harvard Business Review, 90(10), 60–69.
Song, M., Cen, L., Zheng, Z., Fisher, R., Liang, X., Wang, Y., Huisingh, D., 2017. How would big data support societal development and environmental sustainability?
Insights and practices. J. Clean. Prod. 142 (Part 2), 489–500.
Nishant, 2020 et al, ‘Artificial intelligence for sustainability: Challenges, International Journal of Information Management