AI: Is It Worth the Climate Cost?

Technology Policy Brief #113 | By: Mindy Spatt | June 31, 2024

Featured Photo by Indy Silva for U.S. Resist News, 2024

__________________________________

A dire one-sentence warning from the  Center for AI Safety reads “Mitigating the risk of extinction from A.I. should be a global priority alongside other societal-scale risks, such as pandemics and nuclear war.  It was signed by prominent engineers and executives in the field including Sam Altman, chief executive of Open AI, and Dario Amodei, chief executive of Anthropic.  The threat AI poses to the environment wasn’t included, but by all estimates is equally alarming and is already having a huge impact on greenhouse emissions and water use.

Analysis

Google grabbed headlines a few years ago by announcing that it intended to be carbon neutral by 2030.  However, a recent company report shows that goal is further away than ever before; emissions last year were 48% higher than in 2019.  Tech giants Microsoft and Meta have also reported higher emissions and increased water use.  The culprits?  Data centers and the rapid rise of Artificial Intelligence (AI).

With that growth, policymakers and researchers are raising critical concerns about the vast amounts of water and energy AI requires, which is growing by leaps and bounds.  That’s because data centers consume energy in proportion to their computational workload. Training AI models involves continuous and intricate computations on vast databases using increasingly larger hardware setups. All this energy consumption generates heat, hence the need for vast amounts of water to cool things down.

According to the World Economic Forum,  the energy required to power AI is accelerating by up to 36 percent annually. In addition, the “computational power required for sustaining AI’s rise is doubling roughly every 100 days. This means by 2028, AI could be using more power than the entire country of Iceland used in 2021.”

The International Energy Agency estimates the energy use from data centers that power AI will double in just the next two years, reaching 1,000 terawatts, as much energy as the entire country of Japan uses.

These estimates don’t include the additional energy drain of cooling, the amount of which is dependent on where data centers are located; in some locations, it could add as much as 50% to energy use.

The impacts of that energy use also vary.  A report by the nonprofit environmental advocacy organization Friends of the Earth (FOE) predicts that absent major changes AI will “only exacerbate environmental injustice.  Marginalized communities continue to bear the brunt of climate change and fossil fuel production, and studies are already finding that AI’s carbon footprint and local resource use tend to be heavier in regions reliant on fossil fuels.”  FOE’s view is that the risk of misinformation about the climate being created and disseminated by AI is also an urgent concern.

The World Economic Forum and other critics suggest that AI’s negative impacts on the environment can be mitigated by the potential of the technology for innovation in the energy industry including improvements in weather predictions, making smart grids even smarter, and streamlining methods of delivering renewable energy.  These arguments are very similar to the ones made by California utilities during the Smart Meter wars of the 2010s when expensive new digital meters were installed over customers’ objections under the guise that the usage data they produced would help consumers use electricity more efficiently and help utility companies avoid wildfires.  Neither has proven true.

And in the end isn’t the smartest solution, perhaps the only solution to climate change, to reduce emissions?  And the dumbest to increase them?

Alex de Vries, a data scientist in the Netherlands who studies the energy costs of emerging technologies and has critiqued cryptocurrency for its carbon footprint, said in an interview with Scientific American “I think it’s healthy to at least include sustainability when we talk about the risk of AI. When we talk about the potential risk of errors, the unknowns of the black box, or AI discrimination bias, we should be including sustainability as a risk factor as well.”

And he raises the obvious question that other analysts seem to avoid. Do we really need to be using this technology in the first place?

Engagement Resources:

• Friends of the Earth, Artificial Intelligence Threats to Climate Change
• Peter Herweck, Climate Change Won’t Wait for AI — And We Must Not Either, World Economic Forum Annual Meeting Jan 8, 2024
• David Berreby, As Use of A.I. Soars, So Does the Energy and Water It Requires, Feb. 6, 2024

Check out USResistNews.org/AI for more news on Artificial Intelligence policies, technologies, and trends.

Stay in-the-know with the latest updates from our reporters by subscribing to the U.S. Resist News Weekly Newsletter. We depend on support from readers like you to aide in protecting fearless independent journalism, so please consider donating to keep democracy alive today!