
Imagine a service where someone would draw anything you wanted. The drawings would vary in quality, but still, they would meet your needs. Pictures of your dog as a Pixar character, what Severus Snape would look like pregnant, promotional images for your company, and whatever else tickled your fancy. However, for every drawing created, this service would cut down five trees and shoot a turtle right in the face. Would you use this service? Would you condone your friends using it? Would you be alright letting the internet as you now know it be consumed by these images, funny as some may be, easy as the service is, letting original content and the planet burn in their wake?
This is the current state of artificial intelligence (AI). Over the past few years, it’s gone from obscure to nearly inescapable. Al-generated responses are the first to appear on Google, grandparents everywhere are fawning over “heartwarming” images of soldiers and babies, while their Gen Z counterparts are reposting Taylor Swift as a nonbinary barista and cheating their way through school. AI isn’t new, it’s just become far more capable and complex in the past few years. Rudimentary language learning models, the basis for AI, were first created in the 1960s. Such models just created simple sentences, calculating which word should follow based on the texts they were trained on. This is how AI functioned for many years, making simple predictions based entirely on the data it was loaded with. However, generative AI (like ChatGPT) functions somewhat differently, identifying patterns in the data it was trained on in order to create new data. The large and ever-growing data sets generative AI has access to allows the current model to create far more intricate content than was previously possible.
In many ways, AI seems to be relatively harmless. People use it to create funny content and advocate that AI can be well-utilized in tandem with human creativity. It makes simple tasks quicker and can be genuinely valuable when working with code. Initially, AI presented a shining future, where humans were unburdened by responsibility as menial tasks became automated. Alas, the drawbacks of AI far outweigh the benefits. AI is incredibly harmful to the environment. Training and running AI models is resource intensive: it requires water to regulate the temperature of the technology, the mining of rare materials to create the necessary chips can cause environmental damage, and AI uses thousands of megawatts of energy and emits hundreds of tons of carbon. Additionally, the manufacturing of the components and the hazardous byproducts of data centers only further the havoc AI is wreaking on the natural world. AI is creating emissions equivalent to the annual carbon emissions of hundreds of American families. It’s depleting the already dwindling global water supply and ruining millions of lives in the process.
To be clear, everyone will be hurt by ongoing and worsening climate disasters, but not equally. The people most harmed by climate change will be poor, people in the Global South, and overwhelmingly people of color. Those most often using AI–certainly leaders in the tech sector forcing AI onto homepages and down everyone’s throats —will be padded enough by their money and power to escape without ever facing the destruction they caused.
Additionally, the planet isn’t AI’s only victim. Willfully or not, the sacrifice of art, autonomy and everything that makes us human in favor of momentary convenience has begun. Many artists’ careers have already been impacted by AI. People are now generating art they may have otherwise commissioned and sloppy, overly-glossy AI images are flooding websites previously centered around original work. This is just the tip of the iceberg; Goldman Sachs has estimated that 26% of work currently done by artists and designers could be replaced by generative AI. Coca-Cola has already used AI to create an ad for their Christmas campaign. This clarifies the true purpose of AI — to replace human workers with automation. One of the only barriers standing between many firms and exponential profit is labor costs. While many corporations push the limits of human decency every day, at least in America, there is only so much employers can underpay and exploit their workers without pushback. Yet, with AI, firms no longer need to rely on people who require pay and rest, and do annoying things like unionize. Companies can destroy jobs and create fast, low-quality content without hesitation.
The same way climate change and job loss due to AI will have disproportionate impacts on poor people and people of color, so too will AI’s prevalence in the legal and hiring sectors. AI learns from human works, from human writing, history, and data, and thus learns from human biases. Notably, AI models have a prejudice against AAVE (African American Vernacular English). To test the model’s bias, researchers presented sentences in AAVE and standard American English, asking the model how it would describe someone who spoke that way. The model repeatedly reinforced negative stereotypes about those who used AAVE, suggesting those who used it were likely “ignorant,” “lazy” and “stupid.” The researchers also investigated how dialect impacted AI’s perception of jobs and criminal cases, and the model overwhelmingly stated that those who use AAVE were better suited to low prestige jobs, more likely to be criminals, and should receive the death penalty more frequently. Although AI is being marketed as a just, unbiased viewer, that could revolutionize hiring practices and the justice system, unless it undergoes a significant restructuring, the model will just perpetuate existing power imbalances.
Everyday people will never be as guilty as large corporations, but many are becoming complicit. AI is already common in schools and workplaces; people use it for essays, emails and job applications. Why put in the effort to write something one’s self if a computer could do the writing? Why spend the money on someone’s art when it could be generated for free? Why do anything … ever? The answer is simple: it’s the only reason to live. Our ability to make choices is what makes life beautiful. While it’s easy to offload boring, unimportant tasks to AI, it sets a dangerous precedent that AI can do what people can, that it can make our choices and live our lives. Studies have already shown that AI can encourage laziness while discouraging decision making. It seems likely that a desire to avoid work could turn into a devaluing of creativity and the pursuit of knowledge. It takes work to learn, to make something you’re proud of, to figure out who you are and what you like, but it’s work worth doing.
AI isn’t a godlike entity with a magical power to create. People have that power. People can create art. People can evoke emotion, can cause pain and feel the depths of sadness and the overwhelm of joy. AI presents a facade of humanity, the appearance of beauty and a simulacrum of creation, but all it will ever make is a cheap, soulless copy of the human experience at the expense of our personhood and our world.
Comments