Why We Doubt the AI Prophecy
By TMP 43rd Editorial Board
Most articles you will read in this issue will assume that the spread of artificial intelligence into every corner of our daily lives is inevitable. Despite the well-written predictions of our contributors, we seek to challenge this premise: AI is an often-faulty technology whose trajectory we barely understand. As such, we should not place our blind faith in its impending takeover, but rather maintain a critical outlook on its faults when deciding how we implement it.
Before we debate the ethics of AI use, we must first understand the current technical limitations of the technology. AI models “hallucinate”––or, falsify information––at astounding rates. As of May 2025, OpenAI’s o4-mini reasoning system hallucinated at a rate of 48%, per the New York Times, and ChatGPT sometimes seemingly breaks down after simply being corrected. Moreover, while innovation will help refine the accuracy of AI outputs, we still do not understand how AI models make associations based on the data fed to them, and thus, what causes them to be biased; in other words, AI algorithms are essentially “black boxes” into which even the canniest observers cannot fully see.
This faultiness and uncertainty explains why many economists hypothesize an “AI bubble,” wherein the massive investments tech giants have made into AI have overvalued the technology, and that the stock market will crash when consumers realize AI’s technical faults. Microsoft Founder Bill Gates, in an interview with Business Insider, compared this AI bubble to the “dot-com bubble” of the late 1990s, which shaved 78% of the value off the primary US technical stock index.
Even in the face of this uncertainty, our tacit acceptance of the “inevitability” of AI’s expansion has caused many of us to mindlessly and enthusiastically adopt the technology for academic use. Some students insist, for instance, that computer programming and essay writing will become obsolete skills before we even graduate from college, justifying routine academic integrity violations. This attitude fails to recognize that large language models remain incapable of writing to Milton’s standards; should we surrender our learning to an entity that is—for lack of a better phrase—markedly more stupid than we are? A recent MIT study tracking brain activity in writers concluded that ChatGPT users across different age groups exhibited the slowest and lowest brain engagement rates. Undeniably, giving up our brain capacity will not help us adapt to the age of AI, but merely leave us unprepared for the future.
Don’t get us wrong: The Milton Paper believes that AI is here to stay, in some capacity. Tech giants like Meta, Alphabet, and Microsoft will continue pushing AI into classrooms with their investment capital. We are, however, arguing that we can decide how AI shapes our lives. The industry’s growth might indeed be inevitable, but our compulsory usage of the technology is not, so when considering using AI to “think for us,” we must first ask: How factually sound do I need an AI bot to be before I start using it to help with historical research? Why are the arts and humanities still valuable in a world with AI? What skills do I want to get out of school, and how might delegating tasks to AI hinder my learning them? Regardless of anyone’s predictions, when it comes to actually using AI, we must be intentional.
As you continue to read this issue, keep our critiques in mind. Challenge our contributors’ ideas and refuse to take sweeping forecasts as gospel. Keep learning and be critical. Read about AI. Go to AI Club and Programming Club. Open-mindedness is paramount to helping you understand what skills AI will supplant and thus what skills you need to survive in a world where it is widespread. But do this with a critical eye, in particular for the oversimplistic claim that just because AI is on its way, we should simply accept it and adapt as we go. Start debates when necessary, because only by understanding AI’s faults can we adapt to it on our own terms