During his presentation at the Cyborg Futures Workshop (March 31-April 1st, 2017), author Vikram Chandra called attention to an issue previously overlooked during the conference: the presence of bugs and ambiguity in computer code. Other speakers expressed concern over the potential repercussions of using inexact language when speaking about artificial intelligence, particularly in terms of society’s tendency to anthropomorphize robots. Chandra spoke about the effects of ambiguity in more practical terms though, drawing attention to the fact that computer code is a form of communication that is especially susceptible to indeterminate language.
Capitalizing on his interest in both art, which he showed benefits from vagueness, and science, which works to minimize uncertainty, Chandra explored the dualistic nature of the ambiguous by turning to Sanskrit theorists, who attempted to curtail ambiguity while still recognizing beauty in it.
Chandra spoke first of what he called a programmer’s “dream,” a complete clarity of language, allowing perfect communication between man and machine. This is, as Chandra showed, currently impossible, and failures of this dream spawn both computer errors and the message that programmers dread: “an exception has occurred.” But why should communication between man and machine be more difficult than the already challenging task of conversing with other humans? Even between people, words may have multiple meanings and different words can mean the same thing. Context matters, as does setting. Terms can be used literally or figuratively. In essence, coding is a type of formal language, an intermediary between human language and the zeros and ones through which computers operate. The translation is more susceptible to misunderstanding than human to human communication though. As Chandra succinctly explained, computers are dumb. Sentences that are easily understood by people, such as, “Mary ate the salad with spinach from California for lunch on Tuesday” give a computer a multitude of possible meanings. Without the ability to understand context, figurative language, and other idiosyncrasies of human language, the usual pitfalls of communication only increase when we try to speak to a machine.
Ancient Indian scholars recognized the danger involved in misunderstood words. Sacred texts were considered the “code of the universe,” and misinterpretation of these codes could have dire consequences, providing incentive to clarify language. In 500 BCE, a man named Panini created the world’s first generative grammar. Its rules, which were applied sequentially, acted like an algorithm. With an infinite number of possible words, the language was precise, yet flexible, formal and unchanging. It appeared perfectly unambiguous. Ambiguities still managed to exist though. Chandra used the sentence, “the sun has set” as his example. Such a simple sentence seems straightforward, but if spoken from a king to his commanding officer, it could mean that the time has come to launch their attack. If said by a girl waiting for her lover, it could insinuate that the day is done and she still awaits his return. Therefore, while a sentence’s expressed meaning might be obvious, semantics and the power of suggestion can impart uncertainty on the clearest phrases. To counter this ambiguity, scholars developed what Chandra called a “low-level Sanskrit,” a method of writing that specified a precondition, a present state, and the exact means of proceeding from one to the other, similar to the practice of modern coding languages.
While Sanskrit speakers tried to clarify their language, they also recognized beauty in ambiguity through the aforementioned “suggested meaning.” The writing produced through their precise language was exact but dull, resulting in an effort being made to identify what made poetry beautiful. Some Indian scholars suggested it was the incorporation of figurative language, while others believed beauty lay in the style or diction of the verse. Finally, a man named Anandavardhana said that neither denotative nor connotative meaning could account for all the things expressed in a poem. He called the final source of meaning that poets used “suggestion,” and proposed that poetry is made beautiful by the things it does not say. Something as simple as a name can be imbued with enough symbolism and connotation to give multiple meanings at once, which is impossible to do when using words only literally or metaphorically.
Chandra’s talk provided wonderful insight into ambiguity in communication. I appreciated his look at the beauties of ambiguity since it is usually something that is feared in our culture. From coders who want a computer to respond perfectly to their commands, to scientists who meticulously record their experiments to ensure reproducibility, the unknown induces anxiety. Even in Western art, ambiguity often incites unease. A course I took on Romanticism focused heavily on the fear these poets experienced as one of the first groups of writers whose work, thanks to new technology and media forms, would spread beyond their ken or control once released into the world. The anxiety they experienced seems to have focused mainly around an inability to ensure their work was interpreted as intended. Ambiguity could strike, giving phrases unintended meaning, or perhaps turning their words foolish, trite, or offensive. It is always useful to be reminded of the benefits of something one typically tries to avoid, in this case ambiguity, and I appreciated the unique insight that Chandra provided.
Ellen MacIntosh is a student at Saint Mary’s University, Halifax. Her blog post is based on an essay she wrote for English 4556 (Honours Seminar): Animal Life, Social Robots, and Cyborg Futures, taught by Dr. Teresa Heffernan. Ellen was among the students who helped make the Cyborg Futures Workshop a successful event.
(Feature image source: Geek Sublime, by Vikram Chandra, p.111)