It’s hard to avoid seeing any sort of AI-generated content if you’re online at all these days; from ChatGPT helping people write emails to AI-generated images to whole videos being generated. Until recently, there were many tells you could rely on to see what content is created using AI or fake (like how AI could never really do human hands). But now? It’s getting much, much harder, which means figuring out what’s real and what’s not just got much more difficult. And it’s not just the terrifying implications of making misinformation more believable; AI has an increasingly negative physical effect on communities as a whole. From loud data centers that produce immense amounts of pollution and use massive amounts of drinking water to stealing artistic styles and even exact words of entire videos, AI has been running straight into a dystopian nightmare for years.
The data centers that power the various AI systems require a lot of resources, both in the building and powering of the centers. While some companies are taking over existing empty buildings for their data centers, others are building theirs on wetlands and other environmentally important spots. And once built, these data centers are loud and use a lot of water and electricity, enough to cause serious issues for the neighboring communities.
Building AI Data Centers
Amazon has already built seven data centers on a 1,200-acre stretch of farmland in Indiana, with many, many more planned in the same area and in other states. But in building the complex in New Carlisle, Indiana, Amazon has already used a lot of resources and has caused a lot of problems. They’ve hired four general contractors and 4,000 workers to work simultaneously on the site, which does include a 10-acre wetland right in the middle of the complex that they intend to build over. Water and dirt have already been pumped out of some areas to accommodate fiber optic cables and other underground utilities.
These seven data centers are just the beginning for Amazon and yet, community members are already complaining of increased traffic and noise plus some neighboring wells are already dry. And with one permit application stating that Amazon requested to pump 2.2 million gallons of water an hour for two years, it’s easy to wonder where that water will be coming from and to see just how these data centers will be impacting the world around them.
There’s money to be made in the building of these data centers, with most of the money going to the companies behind them. But there are some union workers who’ve travelled across the country and are able to make $2,800 a week because of overtime and bonuses. Eventually, that new construction work will slow and eventually stop, meaning the communities where these data centers are may not have the same need for the many electricians and other tradesmen that they do now. For jobs at completed data centers, nearly half of the states that offer data center subsidies don’t require any jobs to be created and the rest only require a few. The OpenAI datacenter in Abilene, Texas, for example, currently has roughly 1,500 people working on the construction but once completed, the facility will have only 100 full-time employees. And other companies, like Amazon, won’t even employ everyone working at the facilities, as they will contract out lower paying jobs.
Living Near Data Centers
The building of these data centers is just the beginning of the nightmare for communities around them. Memphis, Tennessee has been dealing with the aftereffects of these datacenters since Elon Musk’s AI company, xAI, moved in during the summer of 2024. The company, which powers the social media company X’s AI Grok, has been using an immense amount of the city’s drinking water while also using more than twice the number of methane gas turbines they are permitted to use. This revelation came from satellite photos that the Southern Environmental Law Center took back in April of this year, which shows that there are 35 methane gas burning turbines on the property instead of the 15 they had applied for. Given that the average Chat GPT query requires more than ten times as much electricity as a typical Google search, it’s not surprising that Musk essentially built a power plant that is polluting South Memphis in an attempt to power and then cool his supercomputer.
There are many other communities that are facing situations horrifyingly similar to what Memphis is facing. In addition to air pollution, these data centers contribute noise pollution, with a persistent humming and roaring coming from the facilities 24/7. Folks living near one facility in Virginia have complained that it sounds like a leaf blower that never turns off. Water is another big issue as well, as the computers and servers that run AI tend to run hot and water is used to help them keep cool. Sediment runoff is building up in Georgia, both in homes and streams near datacenters. Homes near completed data centers, like the one Beverly and Jeff Morris live in, deal with contaminated water and a severe loss of water pressure because of the amount of water the neighboring facilities use every single day. There are even data centers in desert areas where water is scarce, like Mesa and Phoenix, Arizona. In a time where the American West is facing major droughts, data centers that use and discard billions of gallons of water every year, it’s not particularly difficult to see why local communities don’t want these centers in their towns.
All of these issues only scratch the surface of the ethical considerations of AI and data centers. For example, a new class action lawsuit is claiming that one AI company, Otter AI, is recording private conversations without permission. Another example is that some folks have been using ChatGPT as a therapist to help solve real life problems, despite the facts that what you share with ChatGPT can be shared with third parties or used against you and that large language models like it are both trained with immense biases and have exhibited high rates of sycophancy (excessive agreement with or flattery of the user). For those in precarious situations like manic episodes, excessive flattery and agreements may sound good but it’s not what’s actually needed.
A very new study has looked at the effects of AI use on brain activity and memory and it showed that using tools like ChatGPT for things like writing essays seems to negatively impact neural connectivity. With such limited research, it’s difficult to make exact statements on how AI impacts our brains, particularly for children whose brains are still developing, but there are those who theorize that an overreliance on AI and related technologies could be negatively affecting how we think and act.
Talking about the negative impacts of AI, like talking about the perils of climate change, sometimes feels like being Henny Penny screaming about the sky falling, especially in the face of how normalized AI has become for an incredible number of bullshit tasks. Even without AI writing silly songs or helping you write a tough email, there can be a lot of good that can come from this technology – like using it to help regenerative agriculture that can produce food in today’s climate crisis or to use it to detect sepsis in hospital patients much earlier and allowing for swift treatment, or to use AI facial recognition to help tract elusive wildlife species from afar.
Unfortunately, there is no grand conclusion for this issue, other than, perhaps, that using AI tools like ChatGPT or Dall-E come at a very serious cost for the environment, your brain and creativity, and local communities around the United States. But while more mundane AI tools will steal from artists, pollute the air and water, and potentially impact your brain, others are saving lives as well. To say that all AI is bad is a grand overstatement but tools like generative AI can be used in ways that are harmful while also damaging the planet. Ultimately, AI should be used wisely and in situations that could actually benefit society and the planet.