AI challenge and promise
A year ago, when the use of artificial intelligence (AI) in the classroom was being discussed and debated, I argued for its use. I was naive to liken AI to the calculator on our smartphones, believing AI was not the enemy, it was a tool that saves us the heavy lifting of daily tasks. Since then, I have been subjecting student papers to plagiarism apps, reading them more than once before grading. In each pass, I was trying to figure out how much of the essay was “assisted” by Grammarly or ChatGPT. In the beginning, I could figure out compositions written by a machine. These used a lot of adjectives, sometimes three in a sentence. Content was mostly long and very descriptive rather than being specific or to the point. Words and certain turns of phrase deployed were not consistent with the compositions of a Filipino college freshman. I began to suspect AI but gave the students the benefit of the doubt. If Mozart composed a symphony at eight, why couldn’t a bright freshman write way better than me? Things got to a point where I jumped and rejoiced when I read a badly written, badly composed essay, because I was sure it was original!
I used to think that AI is only as good as one’s search history. And my search history for Philippine history is far deeper than that of the ordinary freshman student. But then, Google is learning faster than we think. It keeps track of all our online searches and tailors responses based on our digital profiles. It is scary to think that Google knows us better than our family and friends (especially in searches we keep private). Google probably knows us better than we know ourselves. The challenge in academia today is not so much the ethical use of AI but the idea that one need not go to school to learn and get a degree.
I recently came across videos about an app that allows you “to speak with Jose Rizal” and ask him questions. The gullible young people who gushed about it should be reminded that they were not talking to Rizal who died in 1896, they were talking to an AI Rizal fed by everything on him, reliable or not, available online. I decided to play with ChatGPT recently to see how much it knows about specific topics. I asked about Philippine Incunabula, which are extant samples of early printing in the Philippines from 1593 (the year the first two known books were printed in Manila) to 1640. AI told me:
“Before we get into the specific Philippine context, let’s first clarify what incunabula means. The term comes from the Latin word incunabula, which roughly translates to “cradle” or “swaddling clothes.” In the context of books and printing, it refers to those works printed before 1501, when the printing press was still in its infancy. These early printed books were revolutionary at the time, drastically changing how information was disseminated. And while the Philippines wasn’t exactly on the cutting edge of printing technology in the late 15th century, we do have a few early examples that mark significant moments in our intellectual and cultural history.”
As a bibliophile, I have always been interested in early Philippine printing and have read the standard bibliographical works by Wenceslao Retana, Trinidad H. Pardo de Tavera, and Manuel Artigas y Cuerva (none of which are available in English translation). Then there are more recent works on early printing by Regalado Trota Jose and Patricia May Jurilla. Fortunately, these are not online yet so ChatGPT got many things wrong like stating that the 1593 ”Doctrina“ was printed using movable type rather than by xylography. It did point to early printing as “a tool of colonization,” that could be expanded by a student who only uses ChatGPT as a tool rather than making ChatGPT write their research papers for them. ChatGPT did not list down the known Philippine incunabula but took the easy way out by listing “other significant early printed works” like the 1754 “Vocabulario de la lengua tagala” that is not within the dates for incunabula 1593-1640 when it should have referenced the first “Vocabulario de la lengua tagala” published in Pila, Laguna, in 1613. What amazed me was that ChatGPT was able to connect Philippine incunabula to Spanish colonization and developed the essay to conclude with the printing press and the revolution though it referenced the Filipino reformist paper in Spain, “La Solidaridad,” instead of “Kalayaan” published under the direction of Andres Bonifacio or “La Independencia,” the revolutionary newspaper edited by Antonio Luna.
I asked about “Ambeth Ocampo” and sports. ChatGPT replied about a lifelong love for basketball and tennis as “a more reflective pursuit.” I allegedly discovered an orchid during fieldwork in Bukidnon that bears the scientific name “Dendrobium ambethi.” I have a wife named Maria Teresa Ocampo. Children and grandchildren “whose identities remain largely private.” On prompt, ChatGPT composed my obituary, revealing secret lives beyond my wildest imagination. At the moment, I know more than ChatGPT, but it is learning fast. One day it may write my columns and correct my students’ papers.
