AI Thought Leadership Series | Responsible Business Center: Artificial Intelligence and How We Might Use Profit to Justify its Impact on Climate
Responsible Business Center | Aug 28, 2024 | Gabelli School of Business
For all the excitement around the business implications of AI, there are darker thoughts and worries being expressed. For example, former U.K. Prime Minister, Rishi Sunak’s statement that “there is some chance, above zero, that AI will kill us all.” Sam Altman, CEO of OpenAI, and hundreds of top AI scientists, researchers, and business leaders signed a letter one year ago stating that: “mitigating the risk of extinction from AI should be a global priority alongside other societal-scale risks such as pandemics and nuclear war.” But I write today, not as a doomsday scenarist, imagining the techno-apocalypse, a function of the machines becoming sentient, or some such sci-fi, and doing this to us, but rather, that in the reach for profits, we (business, markets, government) may use the prospect of heightened productivity and margins to justify the anti-climate behaviors necessary to facilitate the fullest, most efficient and profitable effect of AI.
We may be about to witness business quietly stepping back from stated Net Zero and other climate objectives. Reviewing recent comments of public companies, they seem only begrudgingly, to make any connection or explain the prospective pivot. However, we merely need to look at public disclosure about the anticipated carbon footprint as well as costs associated with AI (chiefly energy) to see this relationship.
In 2019, Amazon vowed to reach net-zero carbon emissions by 2040. The next year, Microsoft committed to removing more carbon than it emits by 2030. In 2021, Google said it would achieve net-zero emissions by the end of the decade. In recent environmental reports, however, these three tech titans have acknowledged that the race to develop energy-intensive AI features and build new data centers in support, could jeopardize their long-term climate objectives— the tradeoff (they seem willing to make), of progress and profits vs. climate goals.
Microsoft reported its carbon emissions are now 30% higher compared to 2020 as it invests in the infrastructure needed to support AI services. Google also disclosed its carbon emissions have increased by 48% in the past five years due to energy consumption at its data centers, among others. Google warned that expanding AI products will further complicate efforts to reduce those emissions.
These recent disclosures offer some of the clearest indications yet of the potential environmental impact from the economic frenzy around AI.
AI’s Increasing Demand for Energy and Impacts on Climate
- Increased Greenhouse Gas Emissions: The additional energy required for AI operations, particularly in data centers and high-performance computing systems, often relies on fossil fuels. This can lead to higher greenhouse gas emissions, which contribute to global warming and climate change.
- Energy Consumption in Data Centers: AI requires substantial computational power, leading to increased energy consumption in data centers. These centers need both power for running servers and cooling systems to prevent overheating, further driving up energy use and emissions.
- Localized Environmental Impact: The move towards edge computing, where data processing occurs closer to where it is generated, increases energy demands on local devices and infrastructure. This localized increase in energy consumption can strain local power grids and contribute to localized pollution.
Old Fashioned Supply/Demand Calculations – Not So Good
Before we even get to the new demand for power that tech and its data centers will require in order to facilitate expected usage, the North American Electric Reliability Corporation (NERC), the international regulatory agency, projected that a majority of regions in the U.S. and Canada will have insufficient electricity supply to reliably meet demand during extreme weather conditions. More than 300 million people face the growing possibility of electricity shortages beginning as early as 2024 and continuing to 2028. Now let’s consider the anticipated new energy demands related to AI. New AI-related energy demands, systemically not contemplated, are a function of increased data center energy consumption, High-Performance Computing (HPC) systems to power complex calculations and huge datasets, and Edge Computing (processing data closer to where it’s been generated). Some believe that if you were to just integrate large language models, GPT-style models into search engines, it might cost five times as much environmentally as standard search. At current growth rates, some new AI servers could soon use more than 85 terawatt hours of electricity each year, more than some small nations’ annual energy consumption.
But Business as a Force for Good . . .
We still believe! But capitalism as an economic system, may only incorporate people and planet, explicitly, if people and planet are part of the calculus—by choice or by system design (and law). AI is the new toy, the shiny object. Markets love it. This may be the insidious way that “AI kills us”—slowly and by sleight of hand, by hastening the use of fossil fuel-based energy to power AI-stoked profits, as we, employees, shareholders, consumers, hear (and accept?) the carefully crafted two-step on the “Net-Positive Benefits of AI equals Climate Solutions plus Other Social Good minus Fossil Fuel Use ….” as I’ve described in a past writing, the lies we want to hear.
I will say it is possible that with sane governmental oversight and markets and business’ acceptance of it, AI and climate goals have a shot at coexisting. It’s feasible that industry may finance and champion new and renewable energy solutions and AI might help us to unlock and uncover some of these opportunities. More likely or concurrently, business, possibly government, will work carefully to re-engineer the American perspective on nuclear power: an existing source, more accepted abroad. Without careful and collective, intentionally conceived oversight and objectives—considering the prospects for all current and future people, it will be easy for companies and the people that work for them, to look at next quarter’s projections, potentially borrowing forward from future generations. Let’s hope, as we enter a presidential election year in the U.S., that we can get past both the dogmatic coding around the sacred purity of capitalism and the knee-jerk Right vs. Left debates, to pause and consider only Right vs. Wrong.
Written by: Peter Lupoff who is the director of strategy for the Gabelli School of Business Responsible Business Center, which has a mission to inspire, inform, and activate industry professionals, academics, and students to collaboratively foster a more just and sustainable business ecosystem that serves all people and sustains our planet. He also teaches “Impact Investing” at the Gabelli School.
Lupoff is the founder and principal of Lupoff/Stevens Family Office, his family’s vehicle for impact investments, as well as other grant-making, advisory, research, teaching and writing activities.
He was formerly the CEO of Net Impact (2019-22), a 30-year-old nonprofit, 160,000+ member network with a mission to inspire, equip, and activate emerging leaders to make a positive impact for people and planet. He also was CEO of GOOD Institute (2021-2022), which addresses important issues, driving social action and reimagining the possibilities for our shared future by charting a new contract between business and society.