Princeton’s Andlinger Center for Energy and Environment will host their annual energy conference today. At the conference, Microsoft’s Chief Sustainability Officer, Melanie Nakagawa, will discuss the relationship between artificial intelligence (AI) and the energy sector, and moderators and panelists will discuss how AI’s burden on the energy grid can be counterbalanced by AI’s potential to revolutionize our energy system.
These conversations are high-stakes: the success or failure of the country’s efforts to slash carbon emissions and combat the climate crisis depend in large part on the grid. But discussions around AI’s ability to help or hinder the transition to clean energy often ignore the alarming ways that AI reinforces harmful energy infrastructure. Continuing to overlook Big Tech’s complicity in the climate crisis amounts to a new form of greenwashing. This cannot be tolerated.
AI has become the latest technological obsession. Since 2018, when Google’s CEO proclaimed that AI was “something more profound than electricity or fire,” its technological capabilities have grown enormously. Investors, government agencies, and institutions like Princeton have poured money into AI. Proponents of AI argue that the technology is important — if not essential — to tackle the climate crisis. Companies like Microsoft declare that their AI systems will, as Nakagawa herself wrote, “accelerate the deployment of existing sustainability solutions and the development of new ones — faster, cheaper, and better.”
However, while AI could indeed help solve the climate crisis, its total impact on the climate is more complicated than it would first appear. The Andlinger Center conference hopes to explore some of this nuance, but it will likely leave out critical details about the climate harms of AI. The conference’s agenda tells us the kind of story that panelists will share during the event.
To begin, Andlinger’s second panel will discuss the “surging demand for new power supply” that AI is creating. This topic calls attention to the hugely energy-intensive computer warehouses that power AI, known as data centers. Data centers consume a lot of power; some being proposed today use as much power as about 750,000 homes. By the end of this decade, data centers could consume nearly a tenth of the annual electricity production across the country — nearly five times as much as the state of New Jersey uses annually. This expansion is already stressing the grid, as former Federal Energy Regulatory Commissioner Allison Clements may point out, and is causing utilities to propose new fossil fuel power plants across the country. Some utilities have even brought shuttered coal plants back online because of data centers: exactly the opposite direction the nation needs to be moving to tackle the climate crisis.
But, as subsequent panels will demonstrate, AI experts may suggest that these problems are temporary because of AI’s transformative potential. As one afternoon panel’s description reads, “AI can also unlock new approaches for addressing energy and environmental challenges.” Already, it is true that data centers can make significant efficiency increases with upgrades to system design that minimize their grid impact.
In short, these panels appear to suggest that the problems that data centers pose for the energy grid are temporary, but temporary harms will be quickly overtaken by permanent benefits that AI will bring by reducing carbon emissions in the future. This is the story told by supporters of AI.
However, the story is not complete. It leaves out how tech companies have a less altruistic end in mind when they sell their AI technology: enhanced production of fossil fuels. Indeed, these companies are selling custom AI software to the oil industry, raking in profits by making deals that promote and entrench fossil fuel use. This practice undermines the very problems that they claim their AI will solve.
Nakagawa’s company, Microsoft, is arguably one of the worst actors in this scheme. Microsoft executives want to sell AI systems that are tailor made to model oil and gas reservoirs, telling fossil fuel companies that its systems can help improve extraction — maximizing company profits, and also carbon emissions. Microsoft documents obtained by Karen Hao of the Atlantic show that executives hoped to pitch their new GPT software to Chevron to “deliver more business value” — or, put simply, more fossil fuel production. Shell, too, cited purpose-built Microsoft AI software as a tool it can use to optimize its natural gas drilling and processing equipment in its efforts to produce more and more gas.
Deals like this have been common at Microsoft, according to Microsoft’s former senior program manager of sustainability for data centers. The executive left the company because of its climate hypocrisy, explaining that “over the past four years, fossil fuel companies became among the top consumers of Microsoft’s cloud and AI services — and now, generative AI has supercharged the problem.” Microsoft is not the only company making these dirty deals. But it is an industry leader that has pledged to reduce its own carbon footprint and has marketed its AI platform as a way to help others do the same. As a result, its hypocrisy is particularly alarming.
Microsoft’s actions risk setting a dangerous precedent: marketing AI to help the climate problem in the long run, but in the short-term causing a flood of new fossil fuel plants, straying from climate pledges, and selling AI technology to Big Oil. In public, companies like Microsoft like to brand AI as a tool for the energy transition. But in private, they sell it as a way to expand our reliance on fossil fuels. This is a textbook case of greenwashing.
AI may indeed help tackle the climate crisis tomorrow. But today, its technology is used by polluting companies that actively put us into deeper trouble. Those at the Andlinger Center conference have the opportunity to challenge the industry’s greenwashing and call for stronger ethical standards for the use of AI. For instance, they could recommend that companies follow the lead of Google and refuse to sell their AI to the oil industry. When current projections estimate that we are on track for a terrifying 3.1 degrees Celsius of planetary heating, commitments like these are more important than ever. AI’s defenders must change their story, or else they will be tricked by a fiction that hides the reality we face: the world is burning, and fossil fuels must go.
Alex Norbrook is a junior in the History department and an Opinion writer for the ‘Prince.’ He can be reached at an4725[at]princeton.edu.
Read More
Princeton’s non-teaching staff are essential to building community
Asa SantosCommunity extends outside the classroom: Non-teaching employees play a huge part in establishing community and spaces of belonging at the institution.
Community extends outside the classroom: non-teaching employees play a huge part in establishing community and spaces of belonging at the institution.
Re-evaluate your mindset: Put the ‘four essentials’ first
Julianna Lee“We are better at our other obligations when we take care of our four essentials first.”
“We are better at our other obligations when we take care of our four essentials first.”
Why postdocs are demanding open bargaining
Guest Contributors"Princeton’s resistance to the presence of observers is a break from many academic union negotiations. At Harvard, Weill Cornell, Mt Sinai, Columbia University, and the University of Washington, there are no limits on observers."
"Princeton’s resistance to the presence of observers is a break from many academic union negotiations. At Harvard, Weill Cornell, Mt Sinai, Columbia University, and the University of Washington, there are no limits on observers."
