I recently found myself reflecting on a student's written response. The question concerned the medieval saint and philosopher Thomas Aquinas (1225–1274) whose parents disapproved of their son joining the Dominicans and placed him under house arrest in the family castle. The student wrote that while under house arrest, the future theologian and philosopher read the entire Bible. In reality, I had told the students a more legend-like story — that during this period Thomas memorized most of the Bible.

Reading thick books does indeed seem like an achievement nowadays, since summaries and answers are usually just a few keystrokes away. This student's answer, as has often been the case lately, turned my thoughts to artificial intelligence. I asked ChatGPT about Thomas Aquinas's house arrest and received a fairly accurate response:

"While he was under 'house arrest' (in reality imprisoned in his family's castle), he mainly did the following:
Studied and read — especially the Bible and theological texts
Memorized large texts (he had an exceptionally strong memory)
Reflected on his future philosophical and theological ideas
Remained firm in his decision to become a monk, despite family pressure"

There's nothing to complain about — AI outperformed both me and the student by taking into account both the student's claim about reading the Bible and my story about memorization. It even became romantically expressive, recounting the tale of the prostitute hired by Thomas's parents, whom Thomas chased out of his room with a fire poker before marking a cross on the door. The AI wisely added that this is, in fact, a legend.

I am an aging teacher; keeping up with digital matters comes slowly to me and when it comes to AI, I admit my ignorance. Yet even what I have seen so far confirms that AI is more systematic than I am — it creates better slides, video clips and questions for both review and tests. True, the slides sometimes contain rather glaring errors, the clips may feature narration with a strong Finnish accent and AI-generated images can be criticized for their stereotypical nature — but these flaws will likely be corrected in the future.

As a teacher, I have noticed that apt comparisons are helpful for remembering things and for explaining the essence or relationships between ideas. At times, I have begun comparing AI to a kratt — a kind of magical helper from Estonian folklore, which a person constructs from old materials.

AI, too, is something we have made ourselves, containing accumulated knowledge from human history, thoughts that have emerged in societies, as well as analyses and ideas. Like the kratt, AI begins to work for its master, bringing them the property of others — though in this case primarily intellectual property.

However, enlisting AI as a helper is a dangerous undertaking. The kratt is brought to life with the help of the devil, with whom one must make a pact. Typically, this involves giving the devil three drops of blood, which tricksters attempt to substitute with three red berries. The dangers hidden within this symbolic contract with AI have been well described by Jaan Aru in his cautionary writings, especially in his book "Aju vabadus" (Freedom of the Brain) published last year. In such a contract, freedom itself may be lost. I quote from the book:

"Freedom is not doing what I want. Freedom is doing what I want in such a way that it has not been fed to me by someone else. I am free when I better understand who or what led me to that particular choice. This is freedom from algorithms, freedom from goals formulated by others, freedom from obsessions and mental ruts. It is freedom from being under someone's command, but also freedom from hidden manipulation."

As a side note, the excellent artist Kristina Viin, in her painting "At the Crossroads of Four Roads," depicts a young girl holding three red berries in her palms, ready for a dangerous bargain — presumably attempting to deceive the devil. Of course, the painting is open to interpretation and further reflection.

When I asked AI to draw an image of making a pact with the devil, it produced a picture of a sensuous young woman with berries and an equally sensuous horned devil — it was clear they were already in agreement even without signing any contract. It seems that AI does not yet grasp the complex nuances of independent human souls and that encourages one to use one's own brain whenever possible.

Lately, I have been searching for topics where AI is, at least for now, powerless. I recently came across a poem by Nikolai Baturin, "Üü kõllane ku kärjevaha" and posted on Facebook, somewhat mockingly, that AI could not handle something like this. I quote the first stanza of the poem:

Üü kõllane ku kärjevaha
ja lõhnav nagu einäkaar.
Kuu keset luhta satte maha
ku ammustatu valgeklaar.

(The night golden like honeycomb wax
and fragrant like a hayfield.
The moon settles in the middle of the meadow
like a bitten white apple.)

I return to the comparison of AI with the kratt. The kratt constantly demands work can become dangerous if none is given. A foolish master assigns the kratt tasks that reinforce his existing status and, figuratively speaking, such a master may end up living among dozens of bread paddles and flour chests and thousands of smoked hams and beer barrels — sinking ever deeper into his own obsessions. This stands in direct opposition to the ideas quoted from Jaan Aru.

I am still uncertain what a wise master of the kratt of AI should be like. This is important, because I see the presence of this kratt daily in schools as well. A wise master must certainly give the kratt tasks that support the freedom and development of the mind. But as I said at the beginning of the article, I am something of a novice in these matters and therefore grateful to thinkers, writers, officials, teachers and others who patiently and thoughtfully explain to society the nature and use of this useful yet dangerous kratt.

--

Follow ERR News on Facebook and X and never miss an update!