Artificial intelligence has become a national political issue and increasingly a local one. Communities across the country are fighting over the massive data centers needed to power the AI boom. Environmental groups warn about water consumption. Ratepayers worry about electricity bills. Workers fear the technology will undermine their livelihoods.
The backlash crosses ideological lines, and many of these concerns deserve serious attention. But too often, the debate begins and ends with a terrifying number stripped of context. As AI advances, America faces difficult choices. The only way to make them intelligently is to start with facts rather than numbers selected for maximum panic.
Three common criticisms of AI and data centers are frequently overstated. Stories about data-center water use often begin with shocking figures—millions of gallons per day or billions annually. In a water-stressed community, these numbers can matter. But scale matters too: A 2026 analysis by the Florida Water and Pollution Control Operators Association estimated that all U.S. data centers combined accounted for roughly 0.3% to 0.4% of total daily U.S. water withdrawals based on 2021 figures. Cooling methods also vary dramatically—some facilities rely on evaporative cooling, while closed-loop and air-cooled systems use far less water.
Electricity consumption is enormous: Lawrence Berkeley National Laboratory estimated that U.S. data centers consumed about 200 terawatt-hours of electricity in 2025, roughly 4.7% of total national electricity consumption. Critics often assume this translates directly to higher household bills, but recent research suggests the effect on retail rates has been limited. Poor rate design can shift infrastructure costs onto ordinary customers, and badly planned projects can strain local capacity. However, building more abundant, reliable power remains the solution.
Job displacement deserves the most caution. While AI could eliminate certain jobs—particularly entry-level positions—the long-term impact is uncertain. In 2016, AI pioneer Geoffrey Hinton argued that people should stop training to become radiologists because machines would soon outperform humans at reading medical images. A decade later, U.S. radiology has expanded with more doctors and higher salaries. Research shows the number of active radiologists rose over the past decade. This phenomenon—where technology makes a resource or service cheaper and more efficient, leading to increased demand—is known as the Jevons paradox. The same applies broadly: If AI makes programming five times more productive, it might seem fewer programmers would be needed. But economies are not static; new applications become economical, entire product categories emerge.
None of this guarantees every displaced worker finds a new job or that every profession grows. Technological change creates winners and losers. Productivity gains do not automatically translate to equivalent job losses.
The strongest concerns about AI—privacy, surveillance, censorship, individual autonomy, and concentrated power—are real but often overshadowed by sensationalized claims about water, electricity, and jobs. As AI becomes embedded throughout society, these systems could shape what information people see, how institutions make decisions, and how individuals interact with government and business. Massive data collection could strengthen surveillance. Ideological biases might be embedded in algorithms. Dependence on AI could weaken personal autonomy and concentrate decision-making power.
The goal should be to preserve individual liberty, privacy, transparency, personal autonomy, and constitutional principles while allowing innovation to flourish. Exaggerating the easy concerns only makes the difficult ones harder to confront.
Donald Kendal is the director of the Emerging Issues Center at the Heartland Institute.
