Love it or hate it, AI is a hot-button issue right now. For many people, it has become irrevocably embedded into their workflow — from menial nonsense to cutting-edge research, AI has the capability to revolutionize almost every part of everyday life.

I’ve written in the past about the dangers of decrying AI on its face. TL;DR: The response to AI art has been overwhelmingly reactionary. I am constantly surprised by my friends, most of whom would otherwise describe themselves as progressive, fearmongering about “soulless” art and dangerous “misinformation machines,” despite these being arguments used at various points against written language, photography, the internet, image- and video-editing software, digital art, and every other technology that’s threatened the status quo for millennia.

Now, though, times are changing. For example, the number of data centers in the U.S. has skyrocketed from just 311 in 2010 to a staggering 1,240 in 2024, according to Business Insider, and those centers use an equally staggering amount of electricity and water. As a result of companies like Google and Microsoft barrelling into an “AI-first” world, as NPR puts it, global carbon emissions continue to rise exponentially. That’s not to mention the costs of manufacturing infrastructure, building and shipping hardware, paying employees to maintain and repair centers, and dozens of miscellaneous expenses.

But that’s not the whole story. The rise of AI is contributing to increased demand for data storage and processing, yes, but the main culprit is electrification and industrial reshoring, according to an analysis by Goldman Sachs. A commonly quoted statistic purports that a single Chat-GPT query uses nearly 10 times more energy than a Google search. While true, this factoid conceals how much energy both take: the same Goldman Sachs investigation reported that a Chat-GPT search requires 2.9 watt-hours of electricity, whereas a Google search uses 0.3 watts. By contrast, an incandescent lightbulb uses 60 watt-hours on average.

Estimates about data center energy usage also rely on extrapolating current growth into the future. As companies pour billions of dollars into the rapidly-ballooning AI industry, it’s starting to look like the AI bubble is set to burst — and soon. I find it hard to believe that the implementation of AI will continue to grow like it has over the past five years, and I definitely don’t believe investment companies about it.

Additionally, most data centers are used to perform vital tasks that keep the worldwide web up and running, like providing cloud services, hosting websites, and storing necessary data. In fact, data centers worldwide account for only 1-2% of the world’s power usage, and of that, less than 1/5 can be attributed to AI usage. Imagine all the billboards displaying advertisements on empty streets and storefronts lighting empty sidewalks across America. How wasteful is that?

Obviously, then, the issue is not AI usage, but trillion-dollar companies who feel that they can take advantage of the world’s natural resources and the local communities whose lives they have commodified. I have seen first-hand the good that artificial intelligence of all kinds can do: this past summer, I had the privilege of working with a lab at Mass General Hospital that used machine learning and large-language models to detect cancer and predict side effect risks from treatment. Our lab, in addition to proprietary MGH software, utilized the Dana Farber Institute’s version of Chat-GPT. That’s just one example of how AI could catapult the scientific community into a new era of understanding.

Anti-AI stances aren’t free from harm, either. WIRED compiled 22 copyright lawsuits in the U.S. concerning AI training “stealing” intellectual property. Plaintiffs in these copyright cases include such underdogs as Sony, Warner, Getty Images, and Dow Jones. Just last month, a judge authorized a $1.5 billion settlement between Anthropic, an AI start-up founded in 2021, and authors who claimed that Anthropic’s AI chatbot Claude was trained on pirated books, according to AP. The ruling is eerily similar to the lawsuit against the Internet Archive, which aims to make thousands of books accessible online to people who cannot afford them or have no access to a library. They lost the lawsuit in March 2023 to four major publishing companies — another win for authors and readers alike, surely.

If you value art, literature, and freedom of expression, then you should oppose the concept of intellectual property altogether. Stronger copyright laws will only enable companies like Disney to squeeze every last penny out of their employees while keeping a stranglehold on their IPs until the sun explodes. The common refrain that copyright allows small, independent artists to profit from their work is blatantly untrue: has copyright law kept Spotify from paying musicians between $0.001 and $0.005 per stream? Has it prevented Nintendo from sending cease-and-desist letters and DMCA to hundreds of small fan creators? Has it stopped YouTube’s trigger-happy copyright system from being exploited by bad actors?

Trying to legislate artificial intelligence out of existence is a recipe for disaster. There are too many things that will inevitably be caught in the crossfire if we discard AI entirely, including the measurable scientific and industrial benefits of AI. Whether AI is exacerbating issues of environmental catastrophe, copyright overreach, and labor rights, or simply shining a spotlight on them, is unclear — but either way, the underlying conditions must be treated.

Let’s face it: AI is here to stay. It’s now our responsibility to prevent its misuse and reign in the corporations that, if left unchecked, will destroy our planet — with or without AI.