To AI or Not To AI?
Do the hidden environmental and energy costs that accompany AI and tech use outweigh the benefits?
For people working in “high-tech“, it is startlingly easy to remove oneself from the real world. In the digital economy, the source of energy used to power our websites, our computers, and our swanky algorithms, is almost never a cause for concern. In fact, for most businesses, thoughts on servers, hosting, and the like, start and end with the contemplation of buying services offered by Tech Giants like Amazon AWS. The “cost” then translates into a purely monetary consideration. Little do they pause to think about the environmental costs needed to power one tiny site. And therein lies the biggest hurdle that modern digital companies will have to face in order to become sustainable and net-positive enterprises.
I.
There has been a lot of positive momentum geared towards the Circular Economy recently. In a nutshell, the Circular Economy model relies on moving away from the TAKE-MAKE-WASTE linear models into the TAKE-MAKE-REPURPOSE-REUSE cyclical value chains that emphasize high value value extraction and retention. It underscores the importance of studying entire product life cycles and innovating on ways to recover the waste generated at every step in the cycle. The most important and undeniable point made by CE proponents is that our Earth has a finite pool of resources and that it is crucial for us to make the most of what we have.
For companies that work with bio-based products like Cotton and Coffee, it is easier for us to understand and grasp the concepts. For example, we can repurpose cotton fibers into clothing, which can then be made into cushion filling, which can then be shredded and re-spun, or be decomposed back into the earth and made into fertilizer that feeds new cotton plants.
(Full Disclosure: I’m a big fan of the Circular Business Paradigm and may display some bias. )
While reading about companies that are working in this space, I was impressed by how companies like the Plastic Whale from Netherlands (a furniture maker who uses discarded plastic waste) and Oh Scrap, Madras from Chennai (an upcycling clothing retailer) were taking the circular principles and adapting them into suitable business strategies that can simultaneously reduce waste and generate profit for their own companies.
II.
Looking back upon my own work with these new lenses, I was stumped. What about Tech? Where do we even start tracing the environmental impacts involved? As someone who, for a brief period, ran a virtual digital-based company I was extremely interested in figuring out these answers. How much power had I consumed?
Moreover, most of the Autonomous Vehicle companies I had worked for had also consumed tons of electricity, hardware and software resources to make even a small demo of the product available. At that point in time, I hadn’t given much thought to the energy impacts of one demo. But I did see first hand the amount of e-waste that was generated: PCB Boards, cameras, radars and other sensors, and a ton of used batteries, were regularly discarded at the end of their use cycles. Even to my naive eyes, it was clear that the system was missing a way to recycle and reuse these products in a way that is convenient for businesses, the planet and the people.
For this post, I decided to reduce the scope of the problem and try to understand a small portion of it here: What about just software? That should be less destructive, right? After all, it’s one tiny page that opens up in less than a second. How harmful could that be?
As it turns out, VERY. Even a small business that needs only a single lead-generating landing page can use up a good chunk of our planet’s resources.
Here is a simplified version of what it takes to run a small website :-
Through this process, I learnt several difficult truths:
The “Cloud“ is neither intangible nor a wispy harmless part of the world. Our digital world is run on entirely grounded and power consuming physical hardware and transmission lines.
Three main players AWS, Azure and Google, control much of how and where the power comes from. And with the advent of technologies like ChaptGPT and LLMs, they have actually started moving in the opposite direction of their Climate Pledges. 34% more water was used in 2022 by Microsoft, 20% by Google. (Amazon has stuck to its theme of low transparency and has yet to reveal much information on this.) The reality is that the water used by these data centers is usable water taken away from local communities. These tech companies will need to reduce consumption of potable water and instead re-use greywater, etc..
With every passing year, Data Centers are consuming more and more power. And the AI craze conveniently glosses over where this power comes from. An AI powered search can consume more than 10x the energy required by a standard Google search. Countries like Ireland are already facing blackouts and power troubles due to the data centers hosted there.
For AI to be used responsibly, decisions will also need to be made on whether it can be powered by local renewable sources and when/how the transition from fossil fuels happens. Unfortunately, I couldn’t find a lot of public information on the exact steps these companies are taking to counter the environmental impacts of their data centers except that they are focusing more on carbon offsets, which look good on paper, but not so great at actually cleaning up the pipeline.
The good news is that there are smaller players that are trying to build and host green data centers, more efficient transmission cable, and recycle semiconductor components from e-waste. In time, I hope that the pressure from these start-ups, policy makers, climate leaders and local activists will push the Big Tech companies in the right direction as well.
III.
Finally, another timely conundrum we as users and makers need to deal with is this: When should we use AI? What happens when every company starts using complex AI to perform their tasks? Even tasks which don’t necessarily need AI?
The solution may lie in changing how we plan, beginning with the top-down strategy. When companies design products and estimate input costs, it’s important to also factor in energy costs and the carbon impacts. Businesses can and should develop tech that is beneficial for the people as well as the planet. And a major part of that will involve understanding the raw materials cycles used by the tech industry, how and where the energy comes from, and how much of the planet’s resources one is willing to sacrifice for the sake of their product.
Thanks for reading this post! And welcome to my new readers! I appreciate the support, and as always, let me know what you think!