Like it or not, AI has a role in our lives and future. But we can choose how to use it, and whether we allow it to take over our independent thinking and creativity.
We at Shuddhashar FreeVoice are committed to fostering intellectual creativity and promoting diversity of perspectives by amplifying marginalized voices. “To Inspire: Not to Impress” is the motto that motivates our work as a publishing platform. One cannot inspire without being creative first. We, therefore, remain uninspired by syllables strung together without human effort or thoughtfulness. However, as generative artificial intelligence (GenAI) grows exponentially in scale and usage, we must acknowledge that we are on the cusp of something powerful and, for our purposes, deeply troubling. Shuddhashar considers GenAI to be a new, distinct threat to our mission and values, and we urge our readers and writers to consider these risks before heedlessly incorporating GenAI into their artistic processes and ways of life.
When generative AI infiltrates creative spaces, it directly threatens these values by homogenizing cultural and linguistic expression, stunting creators’ artistic processes, and eroding the privacy and livelihood of those we aim to uplift. The AI-powered world brims with correctness that appears impressive, but we believe that creativity and efficiency are generally incompatible—that creative magic often occurs when a writer stares at a blank page, wrestling with some combination of wrong words and right ideas. We recognize that this fruitful process can often be frustrating. In moments of friction, generative AI offers a tempting lifeline: the promise of answers, clarity, and a creative spark. But in truth, GenAI works to smother, not foster, human creativity.
The creative capabilities of GenAI are constrained by AI technology’s reliance on pre-existing information, producing outputs that are superficially novel and structurally recycled. Producing answers that are necessarily derived from training data, GenAI is, by definition, incapable of creating writing that is genuinely new. When users attempt to use this technology to generate “novel” ideas, they are offered synthetic compounds that lack both accuracy and originality.
Meanwhile, as dependency increases, users who rely on Large Language Models (L.L.M.s) for idea-generation are shown to exhibit reduced alpha and theta connectivity in their brains, signifying decreases in creativity and working memory. When users look to machines in the midst of creative struggle—whether to write a novel or an e-mail—their brains become accustomed to navigating around friction instead of through it, and they begin to lose the creative muscle that gives them voice, originality, and ownership. This loss has multidimensional implications, weakening not just our writing abilities, but our capacity for ingenuity, idea-making, and problem-solving—skills that have kept us alive for centuries. Generative AI, by nature, threatens not just what we do, but who we are.
Creativity is not the only risk in a newly AI-powered globe. As an organization built on diverse thinking and writing, we are also concerned about how L.L.M.s undermine language diversity, homogenizing the way humans act, write, and think. Findings from the M.I.T. Media Lab illustrate how AI usage leads to linguistic homogenization, driving human language patterns to become increasingly similar in style and tone. Celebrated by OpenAI founder Sam Altman as a “gentle singularity,” this convergence is biased in nature, shifting outputs toward Western cultural norms and writing styles, and generating anglicized language patterns that reproduce hegemonic power structures and perpetuate harmful stereotypes.
The homogenizing effect of GenAI has particularly harmful consequences for our marginalized creators, as L.L.M.s insist on reducing nuanced narratives to a few bullet points lest they complicate algorithmic averages. A Cornell study recently revealed how Western-developed AI autocomplete tools make biased predictions that change the styles and word choice of Indian users’ outputs to resemble American language patterns, “reinforcing the supremacy of ‘only one language, one culture, one frame of reference.”
This finding signals how, in the absence of diverse data and voices in AI development, subtle, implicit bias—and the cultural erasure that follows it—becomes woven into the fabric of even the most mundane aspects of generative AI. In an effort to produce correct and digestible outputs, GenAI threatens the diversity of perspective upon which knowledge production thrives. In order to preserve the originality and diversity that gives writing meaning, Shuddhashar FeeVoice believes it is imperative that writing remain in human hands.
We can’t ignore the blatant human rights violations upon which most generative AI models have been constructed, and the vulnerable populations and ecosystems they continue to exploit. GenAI is advertised as a universal equalizer: a public resource that promotes equality by increasing access to knowledge. However, reports from Amnesty International and the UN reveal that the creation and implementation of GenAI is incompatible with international human rights law, violating critical UDHR principles that protect citizens’ rights to equality, privacy, safety, expression, and freedom from physical/psychological harm.
Amnesty reports that the training of most commercial L.L.M.s relies on “invasive, mass collection of data” that has been non-consensually, illegally scraped from the web. ChatGPT, for instance, operates almost entirely on scraped data: 60% of training data originates from Common Crawl repositories that hold user information as private as medical reports, whereas 40% is extracted from pre-existing books, scripts, images, and artwork that are utilized without authors’ or artists’ consent.
Similarly, other major L.L.M.s, like Google’s Gemini and Meta’s LLaMa, are also reported to use unlawful data training practices with few options for user protection. Our writers are individuals who often risk their safety to share their voices, despite concerns about how their words and work have been, and will be, used against them. Thus, generative AI’s access to private information—and its ability to release or misuse data, along with its blatant recycling of copyrighted work — directly threatens both our writers’ capacity for artistic expression and their freedom from surveillance as individuals already living under threat.
Contextualized within the industry’s enormous energy requirements, including unregulated water usage, massive greenhouse gas emissions, and construction of data centers in historically marginalized communities, it becomes clear that at-risk individuals will experience the violence of GenAI first and foremost. Yet in the absence of any voices of resistance, this harm can and will spread across the globe, unchecked and intensifying.
We cannot stop the avalanche of artificial generation as it bursts forth into all corners of our joined existence, altering life as we know it—indeed, it already has. Yet, as avid readers of history, we understand that tragedies occur when we collectively step confidently into change, without considering just how dangerous unrestrained “progress” tends to be. As always, Shuddhashar FreeVoice is here to give pause and perspective, offering an alternate way of understanding our evolving and ever-changing globe.
Generative AI would not exist without the human thinking, artistry, and labor it is built upon. Still, if we let machines dominate our creative spheres, unquestioned, we will lose ourselves within the technological dystopia that is already mechanizing and exploiting the most vulnerable among us. We are asking you to help us preserve what makes us, us. We are telling you to ask critical questions. But instead of sending your queries into a chatbot, try asking your neighbor.
Shuddhashar FreeVoice believes in human creativity produced by intellectual labor. We depend, simply put, on you—with your typos and doubts, your original words and honest work, your mistakes and your triumphs, your indisputable, human brilliance. So, send us the words you’ve wrestled with. We can’t wait to see, share, and celebrate them.
You can see Shuddhashar FreeVoice’s Policy on Artificial Intelligence on our Submissions page.
