DIGITAL INSIGHTS: Will AI Become a Net Plus for the Environment?
The question is based on a longer running human debate -- does scarcity lead to devastation, or does it drive innovation?
By Martin Davis
EDITOR-IN-CHIEF
Email Martin
Digital Insights appears on Thursdays and explores the role of data centers in our region. These columns will focus on four areas: tracking the development of data centers in our area, exploring projected and actual tax revenue trends, explaining what data centers are and how they affect our daily lives, and reporting on research and emerging trends in the industry. These columns are made possible, in part, by a grant from Stack Infrastructure.
The future projections about energy use driven by artificial Intelligence (AI) are cause for concern. “In the United States,” according to a report issued earlier this year by the International Energy Agency, “power consumption by data centres is on course to account for almost half of the growth in electricity demand between now and 2030.”
These numbers are intimidating for several reasons — the speed with which this change has come, and the size of growth projected in a compressed timeframe. And they are frequently pointed to as evidence that our environment is on a crash course with environmental degradation.
Though this particular problem is new to the human experience, it’s not the first time that humans have faced equally significant upheavals and wrestled with whether it would cause devastation or prosperity.
A look back is instructive for how we debate and understand the challenge before us.
The Industrial Revolution and Fear of Starvation
At the dawn of the Industrial Revolution in the early 18th century, the concern wasn’t energy use, but food.
Human population growth from the beginning of the Roman Empire to the start of the Industrial Revolution had been modest. In the year 1 A.D., the total human population was about 170 million people. It would take to roughly 800 A.D. for the population to reach 200 million. By 1100 A.D., as Europe warmed, the population reached 300 million. It grew to 400 million around 1450 A.D.
In short, it took a little less than 1,500 years for the population to double.
Between 1700 and 1800 A.D., however — a period of just 100 years — the human population nearly doubled. It grew from 590 million to 1 billion people.
Thomas Malthus was born in 1766, when population was growing at a rate never before seen in world history. Even with the agricultural revolutions being brought by the Industrial Revolution, he felt that farmers would not be able to meet the growing demand for food, leading to mass starvation and a dying off of the human population.
His theory continues to have sway in public discourse, through books like The Population Bomb (1968). Some, like Thomas Robertson who wrote The Malthusian Moment, have tied the environmental movement to Malthusian concerns about scarcity.
Fortunately, to date, Malthus has been proven incorrect. But why have his ideas not become reality?
The answer, it turns out, has to do with people’s ability to solve technological issues.
Ester Boserup was a Danish economist who in 1965 wrote The Conditions of Agricultural Growth, which flipped Malthus’ idea, “arguing that increases in population (or land) pressure trigger the development or use of technologies and management strategies to increase production commensurate with demand.” (For a fuller treatment of Boserup, see this PNAS essay.)
Whereas Malthus focused on scarcity, Boserup focused on the human capacity for adaptation, change, and innovation.
Malthus, Boserup, and AI?
Against this backdrop, let’s return to the original question: Are the energy and water challenges we are facing owing to AI putting our environment on a course to irreparable degradation?
That’s the question that two faculty members at the University of Mary Washington debated in an event put on by the UMW Center for AI and the Liberal Arts early in September.
Arguing that AI would have a net negative effect on the environment over the next decade was Kaitlyn Haynal. Arguing AI would have a net positive impact was Mike Reno.
Haynal based her argument on four points:
AI is built on extractive and colonial supply chains, and it depends on rare-earth metals that are projected to run out.
AI’s production and use are environmentally unsustainable — especially generative models.
AI accelerates fossil fuel expansion.
AI is shaped by ownership and ideology — in other words, it’s created by companies to maximize profit, with no concern for the people or planet.
The argument has Malthusian overtones — notably, AI’s development will ultimately lead to the decimation of the very resources AI needs to thrive and grow.
Reno’s argument, by contrast, has overtones of Boserup. He takes Haynal’s argument point-by-point:
On AI depending on colonialism and extractive minerals, he argues, in part, that there actually are plenty of “rare-earth” minerals, they’re just hard to extract.
On AI’s accelerating fossil fuel production, he argues that the development of renewable energies far outstrips fossil fuels. Further, the administration of AI to manufacturing has already reduced energy consumption.
On AI being based on a reckless “tech-bro”-type optimism, he argues this is unrelated to addressing the resolution.
Finally, on the argument that AI is unsustainable, he states that AI is already reducing fossil energy use and increasing the efficiency of renewables. Further, concerning grid management, AI is critical to balancing the increasing energy needs and the growing renewable energy sources that feed those needs. In fact, without AI, such balancing would not be possible. This is why AI and renewables are “the new power coupled.”
Here, Boserup’s influence is clear. The pressure point — energy use to drive AI — is leading to innovations that will keep us from facing the catastrophic results should energy supplies fail.
The forum itself is more detailed than what is outlined here, but worth the time to get a better sense of the issues and opportunities associated with AI development.
Local Obituaries
To view local obituaries or to send a note to family and loved ones, please visit the link that follows.
Support Award-winning, Locally Focused Journalism
The FXBG Advance cuts through the talking points to deliver both incisive and informative news about the issues, people, and organizations that daily affect your life. And we do it in a multi-partisan format that has no equal in this region. Over the past year, our reporting was:
First to break the story of Stafford Board of Supervisors dismissing a citizen library board member for “misconduct,” without informing the citizen or explaining what the person allegedly did wrong.
First to explain falling water levels in the Rappahannock Canal.
First to detail controversial traffic numbers submitted by Stafford staff on the Buc-ee’s project
Our media group also offers the most-extensive election coverage in the region and regular columnists like:
And our newsroom is led by the most-experienced and most-awarded journalists in the region — Adele Uphaus (Managing Editor and multiple VPA award-winner) and Martin Davis (Editor-in-Chief, 2022 Opinion Writer of the Year in Virginia and more than 25 years reporting from around the country and the world).
For just $8 a month, you can help support top-flight journalism that puts people over policies.
Your contributions 100% support our journalists.
Help us as we continue to grow!
This article is published under Creative Commons license CC BY-NC-ND. It can be distributed for noncommercial purposes and must include the following: “Published with permission by FXBG Advance.”