Two years ago, I predicted an Arms Race (May 2024). See also Another Arms Race (October 2025) and The Arms Race Continues (December 2025). My thoughts were building since AI is a Tool (October 2023).
The challenges are readily apparent to many, but perhaps obscured to some. I recently had a next-gen ask me what I mean by "arms race." Generational differences can hinder communication. The arms race is essentially a cycle in which multiple participants build bigger and better while their opponents do similarly, each striving to outdo the other.
This was a thing in the post-WWII era as the East and West strove economically and militarily for supremacy or at least advantage. Most often, the phrase is applied to the buildup of nuclear arsenals and the threats that presents.
In the world of AI, my thoughts on this were driven by the evolution of AI tools that can make "deep fakes," competing with software that can detect them. But more recently, I have been driven back to the topic by the dreck that is being published.
AI is enabling a cascade of content in social media and news platforms. There are missing words, improper punctuation, and a sea of emdashes. The large language models can prolifically churn out dreck with amazing speed. I have read through volumes of it.
The AI advent has made me cautious about articles and papers. That led me to engage a number of AI and plagiarism detectors that have proliferated on the internet. But that is not where this arms race began. Remember in your youth when you thought "he started it" would gain you some ground?
The current equilibrium results from students making the first move. They elected, years ago, to purchase term papers from professional writers and other vendors. If someone wrote a great paper on Einstein at East West University in 1972, someone else could readily turn the same paper in at West East University in 1973.
Papers and more were traded across the paradigm we called "mail," a process that involved mailing a check to some dark basement, where a quiet troglodyte would cash the check and mail a paper back to the customer. The smart customers retyped those papers. As I conclude that paragraph, I have gone back and added links so that all the antiquated terms can be accessed easily for clarity.
The danger was two-fold:
- First, the professor could tell a paper was way too articulate, logical, organized, etc. for a particular student.
- Second, the troglodytes did a poor job of making sure that only one West East student received each paper.
Spreading and sharing information was hard in those days. It required copy machines, postage, paper, and mail. Then came the halcyon days of the internet. Digital files could be sent in an instant. If a student needed a great paper on quantum finance, they no longer needed to order a month in advance. And the plagiarism flourished.
Companies popped up. They allowed teachers to scan and submit papers for plagiarism long before the advent of PDF and the simple plagiarism filters of today. As a paper is turned in today, it is likely to pass through a filter automatically and arrive at the professor's desktop complete with a report on what may or may not have been pilfered from someone else's thoughts.
Companies began offering students the chance to submit their papers to an analysis platform first ("check it first" platforms). They proclaim that what they detect is "exactly what your professor will see from _________," (insert submission filter product).
As the world advanced, along came large language models (LLMs) in which artificial intelligence saves the student all those arduous minutes of cutting and pasting from the internet to create a paper. The plagiarism is built into the LLM output. All the student has to do is ask the LLM to write a 500-word paper on quantum finance, complete with a bibliography, and within 30 to 90 seconds, the whole product is delivered.
But what of the education industry? Well, companies developed AI detectors to go with the previous plagiarism detectors. And the "check it first" platforms evolved to offering to pre-detect that also. But they added a new feature. They will detect for free, but for a few dollars and a single click, these will "rewrite your paper to eliminate AI and plagiarism risks."
No doubt that the universities will soon have a tool to detect papers that have been "rewritten to eliminate." And the cycle will persist, and every response of a student will be answered with a tool for the teacher, and vice versa.
The real benefit will be less of the nuisance teaching and learning (sarcasm, apologies). As more and more time is devoted to cheating or catching cheaters more effectively and efficiently, there will be less time to teach or learn quantum finance. We will succumb societally to the great dumbing of America.
Generations will falter. Workplaces will suffer. People will fail to produce. Their employers will struggle to thrive. And it is all for the sake of cheating and the mindset that exalts being a better cheater over writing your own paper, learning, and growing.
I am struggling not to be the old man screaming at the kids. But in all honesty, it is getting harder every day. Write your own paper, for goodness' sake.
