Advocates highlight the benefits of AI, such as enabling doctors to make more accurate diagnoses and helping scientists develop new technologies that lead to better, more affordable batteries. These advantages are significant and should not be dismissed.
However, a growing number of critics worry that AI could have long-term negative consequences. Paul Krugman recently discussed with Heather Cox Richardson how the current enthusiasm for AI resembles past economic bubbles - from Amsterdam’s tulip mania in 1636, to Bernie Madoff, the dot-com crash, the California gold rush, and the 2007 global financial crisis.
The frequency of these events suggests there is something inherent in human nature that leads to boom-and-bust cycles. Krugman references economist Robert Shiller, who described a bubble as a natural Ponzi scheme: people profit because others join in, but eventually, the system collapses when there are no new participants. In the end, the money is not truly there, and the cycle depends on a constant influx of newcomers.
Opposition to AI is increasing. Data centers, essential for AI, have become a contentious issue in US politics. Recently, Utah approved a massive data center that, according to Grist, will consume 9 GW of power - twice the state’s current usage. Spanning 40,000 acres, or twice the size of Manhattan, it is expected to raise Utah’s carbon emissions by 64 percent. Grist also noted that the center’s water needs are unknown, but it will be located near the shrinking Great Salt Lake, which is already at risk due to drought.
Robert Davies, a physics professor at Utah State University, warns that the data center could create a massive heat island, devastating the local ecology. He estimates the project will cover as many square miles as Washington, D.C., making it the world’s largest data center, and could raise nighttime temperatures by up to 28 degrees Fahrenheit in the high desert valley. “What I’ve found is, it’s so much worse than I even thought it would be,” Davies said.
Davies compared the additional thermal load to “about 23 atom bombs’ worth of energy dumped into this local environment every single day,” warning that such heat could devastate the area’s ecology. He questioned the impact of depositing so much energy into a fragile watershed at the north end of the Great Salt Lake, a high-desert valley already in crisis. The answers, he said, are deeply troubling.
In New Jersey, the Millville Board of Commissioners voted to ban data centers in the city, passing an ordinance stating that data centers are incompatible with the city’s land use, infrastructure, and community character. The ordinance concluded that data centers would be detrimental to public health, safety, and welfare.
This decision halts the proposed 1.4 gigawatt Millville Energy & Data Center Campus, which would have covered more than 60 acres. The Climate Revolution Action Network, an environmental nonprofit, organized residents to oppose the project. “This is a winning coalition and something we need to see more of across the country,” said Kayleigh Henry, one of the group’s leaders. “These corporations may have more money than us, but they’re no match for people speaking out and making their voices heard.”
This month, a coalition of anti-data center groups asked New Jersey Governor Mikie Sherrill to impose a moratorium on new data centers using at least 20 megawatts of power until regulations are in place to protect consumers, maintain grid reliability, and minimize environmental impacts.
The Memphis Flyer reports that Elon Musk’s xAI data center in Memphis, Tennessee, purchased over 25 million gallons of water from Memphis Light, Gas, and Water in March to cool its Colossus 1 data center. Colossus 1 is just one of several xAI data centers in the area, with others located in Mississippi.
Protect Our Aquifer noted that xAI’s water usage now totals 812,502 gallons pumped daily from the Memphis Sand Aquifer, with expectations that demand will rise in the summer. “If companies are making multi-billion dollar commitments tied to xAI’s facility, then Memphis residents deserve transparency and public accountability,” the group stated. It was also revealed that Anthropic has agreed to pay xAI $1.5 billion a month through 2029 for use of its computers.
MLGW CEO Doug McGowen told the Memphis City Council that Musk’s company was “committed” to building a wastewater treatment facility for its supercomputer, but the company later said it would prioritize other projects and continue using aquifer water. The project’s cost has risen from $80 million to around $200 million. Local residents are frustrated that xAI pays 19 cents per 100 gallons of water, nearly half the 32 cents regular customers pay.
Water concerns add to worries about emissions from the 50 portable methane-fired generators powering the xAI data centers. Tennessee and Mississippi officials have dismissed local complaints about pollution. Environmental justice in the South remains limited, with emissions concentrated in low-income, predominantly minority neighborhoods that have recently faced disenfranchisement by the US Supreme Court.
Questions are also being raised about the necessity of so many data centers and the expansion of artificial intelligence, especially as concerns about actual intelligence grow. Germany’s DW recently published an exposé on how AI networks in China enable extensive social control, a development that raises significant alarm.
A computer expert known as NetAskari told DW that Chinese authorities no longer rely solely on police cameras. Their AI-enabled system can track a target’s specific train carriage and seat, synchronizing facial recognition data from ticket gates at ski resorts directly into its tracking system. NetAskari accessed a site where the system identified his recent ski companions, mapping their movements in detail.
The goal, he explained, is to process as much data as possible from numerous sensors in real time. The system logs daily behaviors such as gasoline consumption, shopping locations, and visits to areas of civil unrest. This massive data fusion creates a comprehensive “holistic personnel archive” that tracks physical movements, consumption habits, and digital footprints.
Previously, foreign reporters in sensitive regions like Xinjiang could evade plainclothes police, but algorithmic upgrades have made such tactics obsolete. “They don’t need to send two or three cars to follow you anymore,” NetAskari said.
With access to mobile payments, ticket purchases, and social networks, authorities can anticipate a person’s itinerary and control what they see upon arrival. If the system detects contact with certain individuals, police can intimidate sources behind the scenes. In this closed surveillance loop, the possibility of an “under-the-radar investigation” is disappearing.
This shift poses challenges for authors of spy novels, as surveillance now tracks individuals openly and precisely. In the Chinese system, people are reduced to numbers, patterns, and vectors - a “data mass” that can be controlled, shaped, and coerced as needed.
.svg)
