Beyond the Catastrophizing: The Importance of Critically Engaging with AI in Modern Education

Keith Madison ’24

Contributing writer

Nearly three in five college instructors report that they or their students are already using AI-powered technologies in their classrooms — this coming from a 2023 survey of more than 2,000 educators across North America conducted by academic publishing company Wiley. Emblematic of a surge in popular discourse surrounding recent and remarkable advances in so-called “large language model” technology (the innovation behind such “conversational chatbots” as OpenAI’s ChatGPT, a computer program designed to simulate human conversation by mimicking the linguistic form of natural language), this statistic and others like it are seen by some academics as cause for alarm. With experts and commentators worldwide warning of impending apocalypse — from widespread academic plagiarism to cyberpunk wastelands — it might seem as though these concerns are well-warranted. Our forthcoming digital overlords should keep us up at night, right?

As someone who builds these tools and has spent several years conducting AI-adjacent research, the ever more creative doomsday scenarios are to me but a humorous distraction from the real, exciting and far less ostentatious work going on in the field (I am particularly fond of one Eli Yudkowsky’s prediction of imminent extinction). Indeed, any cataclysmist who has spent more than a few minutes with one of these chatbots should be quickly put at ease. A recent “conversation” between myself and one such chatbot:

How do you spell “sandwich”?

“H-A-M-S-I-Y-W-I-C-H-M-M-C-H-E-E-S-Y-F-A-C-E”

Terrifying indeed.

Disconcertingly, however, it has become increasingly clear to me that these gross mischaracterizations are symptomatic of something far more acute: a lapse in critical thinking and basic technological literacy of a kind that is particularly threatening to the ethos of a liberal arts institution such as ours. Indeed, many of the loudest voices ostensibly haven’t even taken the time to use these things, much less to familiarize themselves with the machinery behind them, its affordances and its limitations. How can we hope to educate students on the real, present moral and ethical quandaries posed by these tools and inform constructive institutional policy if we have no shared, basic cognizance?

The technology behind today’s state-of-the-art AI systems is neither magic nor fundamentally inscrutable. When you or I present a query to an AI chatbot, the system powering it forms a response word-by-word by conducting a statistical analysis— not unlike the predictive text feature available on most modern smartphones. To do so, it analyzes enormous corpora of human text, “identifying” what words are most likely to follow other words, phrases and linguistic structures. Accomplishing this feat requires untold amounts of manpower, comprising millions of (grossly underpaid) human crowd workers across the globe who are tasked with systematizing these datasets. The result is a tool that is capable of producing a likeness of human language, but which importantly lacks anything approaching a facsimile of human understanding. Moreover, these chatbots cannot achieve the full range of human originality, for they are limited to churning out only some permutation of what we humans have already fed them. Each of these constraints is imposed by the very nature of the underlying technology — one that has remained fundamentally unchanged for decades — which is to say that AI chatbots are decidedly not poised to upend modern education or otherwise facilitate widespread chaos in the near future. To suggest, for instance, that any college assignment could be satisfactorily completed with modern AI is to suggest it requires no nuanced reasoning or understanding. In short, the AI apocalypse is not inbound.

Here now, however, is an unprecedented evolution in the ways we produce media and automate prosaic labor that is simultaneously capable of supplementing the learning process and wrought with far more tangible ethical quandaries. Students are using these tools today in facilitating material comprehension, proofreading, revision and ideation quickly, affordably, and in many ways more accessibly than ever before. Those same students will be engaging with these tools tomorrow as they enter an increasingly dynamic workforce wherein occupations as diverse as customer service, freight transportation and financial advising becoming progressively augmented or even automated away by AI. Yet, because the data that powers today’s AI is scraped from the internet, it is saturated with errors, biases and heuristics that are representative of the hegemonic view characteristic of the loudest voices on digital platforms — predominantly young, white, heteronormative and moderately affluent men (according to one of many such studies conducted by researchers at Buffalo State College). In spite of corporation’s attempts to implement “hard-coded” guardrails, it remains relatively easy to (intentionally or inadvertently) coax ChatGPT and its ilk into producing abhorrent misinformation and revealing encoded biases. Helping students to critically assess these affordances and limitations is vital and requires that we thoughtfully engage with the technology.

The role of a liberal arts institution is to impart its students with a worldly education and a critical eye. This is reflected in our college mission statement, the rousing murals that adorn campus and the presidential address that precedes each convocation. AI holds a legitimate presence in the world, and we as a liberal arts institution cannot confidently graduate students into that world whom are unequipped to make use of and critically assess its promise. Doing so requires that we act now to engage with these extraordinary tools, talking the cataclysmists down such that we may begin to critically reflect on their role in modern education. This is not a task that can be accomplished from atop some pedestal, but one that will require today’s faculty to immerse themselves in it. If you haven’t yet tried these things, you can’t understand the very practical concerns that actually keep those of us who have up at night.