Artificial intelligence is becoming ubiquitous—ChatGPT, Copilot, and DeepSeek get hundreds of thousands of visitors daily. AI art is omnipresent on sites like Pinterest and Instagram, and enthusiasts claim that even mediums like music videos and films will be entirely created by artificial intelligence programs in the not-so-distant future. Individuals who need advice or a question answered no longer flock to Google and look through hundreds of results; instead, they get a succinctly written paragraph that may or may not be factually correct. Some have even used it for therapy and advice by divulging their issues and hoping for the coding behind the chat bubble to have something useful to say.

At first glance, it seems ideal—the stuff of futuristic movies we’ve all seen at least a few of growing up, a true manifestation of what the 2020s were expected to be. Having an all-knowing robot in your palm seems mere steps away from flying cars—and arguably more convenient, too. The generative AI we’ve been served up in the past few years, however, has a much darker side to it—how it works and what it truly costs the people who use it.

Workers as machines

When describing generative AI to someone who isn’t in the know, a common response is: “Isn’t everything on the internet AI?” However, the definition of what it means has shifted as generative artificial intelligence becomes more accessible. What is currently most commonly described as AI is based on machine learning through artificial neural networks, requiring immense amounts of data to be provided for the programs to “learn” from.

And, naturally, there are people behind providing this data: data laborers and content moderators being just two of dozens of positions whose responsibility it is to essentially teach AI programs what to do and how to do it for far below minimum wage.

To put it simply, the currently growing AI industry could not run without the exploitation of workers from vulnerable communities who are offered no protections or decent working conditions. Hiring people through third-party companies and taking them on as contractors to avoid providing benefits is also common practice—all to provide the same function as Google but in chat form. The question of whether it’s worth it shouldn’t even be posed.

Environmental effects

The environmental effects of AI, specifically generative AI, are varied: it begins with the immense amounts of power and electricity and the resulting CO₂ emissions necessary to train and run the systems in question.

The data centers that house AI deployments further contribute to the environmental detriment, taking the issues even further than already fatal greenhouse gas emissions. The elements necessary to create the microchips that power the programs, for one, are often mined in environmentally destructive ways.

It, of course, doesn’t stop there: waste produced by the data centers in question regularly contains dangerous substances, such as lead and mercury. Additionally, and perhaps most controversially, the centers utilize water to power and cool electrical components, using six times more water than Denmark (population: six million) does, according to UNEP.

While still attempting to get a handle on other factors detrimental to the environment, it is arguably impossible to balance trying to fix the environmental damage already done with the damage incoming due to the increasing use of data centers to power AI.

Cognitive decline

The feeling of having a magic, question-answering genie in the palm of your hand has also, expectedly, had negative effects on cognitive development, learning, and decline. Students are increasingly depending on AI for assignments and answers to questions instead of looking to libraries for books and journals. Adults with office jobs turn to ChatGPT for writing simple, straightforward email responses. Of course, this eases the process of getting through university or the workday, but it also causes issues with problem-solving abilities, comprehension, and analytical skills.

Consistently looking to an AI bot to simplify concepts means the development of individual cognitive skills is completely unnecessary to function. Why go to the trouble of research, reading, or even asking someone a simple question when the robot in your phone has all the answers?

The rising, omnipresent lack of comprehension means a rise in illiteracy as kids and students accept whatever information an AI bot provides without truly understanding the meaning behind the concept at hand. Generations with generative AI readily available are likely to have a cognitive function collectively weaker than has been recorded.

With the negative environmental effects in question, the ramifications it’s proving to have on the people fueling it, and the societal effects becoming increasingly hard to ignore, it is possible that generative AI is on its way to having detrimental contributions to several aspects of our day-to-day lives. Of course, there are benefits to having technological advancements at the ready anytime—making work lives easier, having wider access to information, and the ability to research and learn, to name a few—but the cost of generative AI simplifying already simple things may already be too high, and it’s just getting started.

References

United Nations Environment Programme. (2024, September 21). AI has an environmental problem. Here’s what the world can do about that.
Williams, A., Miceli, M., & Gebru, T. (2022, October 13). The exploited labor behind artificial intelligence. Noema Magazine.
Westfall, C. (2024, December 18). The dark side of AI: Tracking the decline of human cognitive skills. Forbes.