The Machine-Translated Web: Signs of a New Educational Reality

In 1995, the renowned educator and media ecologist Neil Postman wrote: "Technological change is not additive; it is ecological. A new technology does not merely add something; it changes everything."^Footnote1 Thirty years ago, Postman was facing the rise of the internet and its inevitable integration into the classroom and daily life. Not since the invention of the printing press had there been a moment in technological history with such transformative potential. So it's important to heed his insights at the onset of this AI era, which, by all accounts, is every bit as significant as the emergence of the internet age.

Last September, I wrote an Inside Higher Ed opinion piece that discussed the potential of current generative AI models to produce counterfeit knowledge en masse. I argued that such an inundation would degrade "our faith in quality work, truth, educational institutions, and even the written word itself."^Footnote2 For some, this may have seemed like hyperbole. After all, these AI models are a relatively new technology, with ChatGPT being just over one year old. However, a paper released in recent weeks by a group of researchers from Amazon and UC Santa Barbara indicates that such an inundation has already occurred, raising serious concerns on societal and technological fronts.

Rise of the Machine (Translator)

In the eye-opening paper, the researchers uncover a striking aspect of the digital landscape: a significant fraction of the web, particularly content written in less commonly spoken languages, has been generated not by humans but through machine (AI) translation. For this study, the researchers created the most extensive multi-way parallel corpus to date, comprising 6.4 billion unique sentences in 90 languages, and found that 57.1 percent of these sentences resulted from AI translation into multiple languages.^Footnote3

This excessive use of AI translation raises concerns about the quality of information online. The study reveals that such content is often of lower quality and exhibits biases in topic selection, favoring the shorter and more predictable, predominantly from general topics like conversations, opinions, and listicles intended to drive ad revenue. The research highlights the need for critical scrutiny of web content, especially in educational and research contexts, as the authenticity and integrity of information in low-resource languages (especially from Africa and other areas in the Global South) are already compromised by this pervasive and cavalier use of AI translation.

Why This Matters to Educators

The role of educators extends beyond the mere transmission of information. As the keepers of intellectual integrity, educators are responsible for preserving and disseminating knowledge. In this era of rapid technological advancement, when generative AI is reshaping the fabric of information dissemination, educators must critically examine how these technologies alter our perception and understanding of the world. The educator's profession carries with it a profound responsibility and an implicit trust to safeguard the quality and authenticity of knowledge. This safeguarding becomes increasingly crucial as the world confronts challenges like those highlighted in the research noted above. It's a daunting task, but engaging proactively in the evolving dialogue is essential. With this in mind, I'd like to offer three critical areas that educators need to address: (1) the effects of this new technology on academic honesty and language learning, (2) the potential erosion of academic discourse, and (3) the need for AI literacy among educators and students.

Academic Honesty and Language Learning

Thirty years ago, as the internet started to make its way into homes, educators grappled with concerns similar to those raised by AI tools like ChatGPT. The ease of accessing and plagiarizing online information reflects today's worries that AI-generated work will mask students' actual abilities. In response, academia evolved: plagiarism-detection tools became commonplace, and citation standards were adapted to uphold academic integrity. This adaptation reflects the continual need for educational systems to evolve alongside technology, ensuring that innovations bolster rather than compromise educational standards. Educators not only must remain actively involved in the ongoing discourse about the role of technologies in education but also must possess a fundamental understanding of how these technologies function, enabling educators to contribute effectively to discussions around solutions and adaptations.

For language teachers like myself, another subtler concern is hinted at in the machine translation research paper. The widespread use of AI translations circumvents the need for the essential rigors of language learning. Particularly for low-resource languages, this risks creating a homogenized version of the languages, one devoid of the subtleties and richness essential for true linguistic and cultural understanding. This is a disservice to students who seek to learn languages in their authentic form and to those for whom the target language is an essential element of their culture and identity.

Erosion of Academic Discourse

Another immediate concern is the potential erosion of faith in academic discourse. Our uncritical adoption of social media, a technology that is only around twenty-seven years old, has profoundly impacted how we see and interact with our world. And a significant part of this impact, we have come to realize, is dangerous and detrimental. As social media has skewed our trust in the authenticity of what we see and hear, generative AI opens the door for a deluge of counterfeit knowledge into our social feeds and academic discourse. This technological evolution is a sea change, reshaping our epistemology—that is, how we acquire and trust knowledge. In my article for Inside Higher Ed, I discussed how these technologies could distort our understanding of truth, mirroring how social media has sometimes degraded public discourse. Educators face the urgent task of navigating this shifting landscape, where the distinction between genuine knowledge and artificially generated content becomes increasingly blurred. This is not just a challenge to academic integrity but a fundamental shift in the epistemological foundations upon which education and scholarly discourse rest.

AI Literacy

It's reasonable to conclude that we have irreversibly crossed into a new era with the widespread adoption of today's generative AI tools. This new reality, already reshaping our world, cannot be ignored or rejected by educators. The significant changes outlined in the reported findings on AI translation are just the beginning. Returning to Postman's words, we don't merely coexist with AI; we inhabit a world fundamentally altered by it. Understanding AI—its mechanisms, capabilities, and implications—is now a prerequisite for effectively contributing to shaping our evolving reality. We must cultivate AI literacy, going beyond the surface-level hype and hysteria to truly grasp the historical context and the nuances of this change.^Footnote4 This knowledge is not just about keeping pace with technological advancements; it's about maintaining our agency in guiding the future.

*****

Despite the dramatic shifts brought by AI, educators' core role remains unchanged: shaping the future. Educators have always been, and continue to be, pivotal in influencing what's to come. For all of us involved in education, our commitment to vigilance and our investment in education must remain constant—even in an era that will change everything.

Notes

1. Neil Postman, The End of Education: Redefining the Value of School (New York: Knopf, 1995), 192. Jump back to footnote 1 in the text.↩
2. A. G. Elrod, "'On Bullshit' and AI," Inside Higher Ed, September 14, 2023. Jump back to footnote 2 in the text.↩
3. Brian Thompson et al., "A Shocking Amount of the Web Is Machine Translated: Insights from Multi-Way Parallelism," arXiv, January 11, 2024. Jump back to footnote 3 in the text.↩
4. Alyson Klein, "AI Literacy, Explained," EducationWeek, May 10, 2023. Jump back to footnote 4 in the text.↩

A. G. Elrod is Lecturer of English and AI Ethics at HZ University of Applied Sciences in the Netherlands.