Sitting in a class about AI, machine learning, and algorithms, I never imagined I would encounter a concept reminiscent of something I had read long ago during my high school years.
I still remember the opening words of the story "The Garden of Forking Paths" from Jorge Luis Borges' collection Ficciones. It tells the tale of Yu Tsun, a Chinese spy working for the Germans during the First World War. His mission leads him to an enigmatic labyrinth, a creation of his ancestor, which turns out to be not a physical maze but a novel.
As the professor moved through the slides, explaining how classification methods and decision trees work, I couldn't help but draw parallels to Borges' writings. Could Borges have anticipated the concept of decision trees as a literary device long before it appeared in the realm of computer science?
In Borges' story (spoiler alert), Yu Tsun faces the dilemma of assassinating a man named Albert. This act serves as a coded message to the German forces to initiate a bombing. Yu Tsun, however, is perpetually confronted with choices—mirroring a labyrinth that is essentially the book his ancestor wrote. Each decision he makes leads to a different outcome, much like the nodes and branches of a decision tree.
While attending the lecture and taking notes, I pondered whether Borges ever considered a career in computational science, or if he was simply an extraordinarily imaginative writer. It is intriguing to think that Borges may have conceptualized "decision trees" long before the field of computer science formalized them. Indeed, many physicists have cited Borges in their work, referencing his ideas on quantum mechanics and other topics seemingly unrelated to literature. Borges himself once humorously remarked upon learning this, saying, "How imaginative physicists are."
The parallel between Borges' labyrinth and decision trees in AI is striking. Both involve navigating a complex network of choices, where each decision leads to further possibilities. This reflection inspired me to write this article.
Decision trees, a fundamental component of machine learning, are essentially a model of decisions and their possible consequences, including chance event outcomes, resource costs, and utility. They provide a map of the various paths one might take, similar to the bifurcating paths in Borges' story.
In "The Garden of Forking Paths," every decision opens up new, divergent possibilities, creating a myriad of potential futures. This is akin to how decision trees in machine learning split at each node, representing a choice that leads to further branches. Each path, or branch, leads to a different outcome, just as Yu Tsun's choices lead to various narrative possibilities within Borges' labyrinthine story.
Perhaps Borges intuitively grasped the nature of complex decision-making long before the advent of modern computational theories. As for me, I still have much to learn to ace my upcoming exam. Maybe I should revisit Borges' work from time to time for inspiration.