Of Prompts and Paranoia
Artificial intelligence and anti-intellectualism — two AIs — share a fascinating dialectic. With the burgeoning of AI, anti-intellectualism has found a new and insidious alibi. Instead of critically engaging with a text, people conveniently ascribe it to artificial intelligence.
The rising tide of AI chatbots, with ChatGPT leading the charge, will not abate anytime soon. Ever since their launch, they have been accepted with rapacious enthusiasm. Students began using these chatbots to write their assignments. Professionals across the board followed suit. The allure of this technology was beguiling: it spared people the ‘ordeal’ of thinking, of having to ‘drudgingly’ pen their thoughts themselves. It was as though a magic genie had been discovered, which could conjure up an entire write-up out of thin air in a couple of seconds — something that would take human writers hours, and for the lackadaisical ones like myself, days.
However, a few things should be clarified at the outset. These chatbots — at least for now — don’t have any consciousness. Notwithstanding the technological advances, as Ted Chiang demonstrates in his article entitled ‘ChatGPT Is a Blurry JPEG of the Web’, what we still get is a generic rehash of the material already available on the Internet. Thus, they come in particularly handy for the ‘content mills’ to avoid copyright infringement, but, apart from that, they serve no real purpose. Even the content generated is, quite frankly, tasteless and banal. Naturally, over time, people started observing certain patterns in the content generated by AI. Here, things started to turn murky. The euphoria generated by this avant-garde technology was eclipsed by paranoia. People grew suspicious of all prose that tended to be embellished and deployed the em dash. Indeed, one of the aims of this write-up is to attempt to salvage the em dash, for many writers have stopped using the em dash lest their writing be falsely flagged as AI-generated.
On LinkedIn and Twitter, my feed is inundated with people claiming that the em dash is a sure marker of a text generated using AI. Some people have gone as far as to claim that ‘no human uses the em dash’. Whether or not AI is predisposed to overusing the em dash is a separate issue, but to attribute a text to AI simply because of a punctuation mark is preposterous. It’s a truism — a truism that unfortunately requires reiteration — that people have been using the em dash long before AI chatbots even existed. I don’t remember reading a single book where the author had not sporadically used it.
The claim that no human uses the em dash is not only patently untrue but also deeply offensive, for many writers tend to use it quite generously — to the extent that it does sometimes appear overdone. It also begs the question: if humans never used the em dash, how come AI magically started using it? AI certainly didn’t concoct a new punctuation mark. It has been trained based on data procured from articles and books written by humans, and that is precisely where it picked up the em dash.
People are free not to like a punctuation mark; they can have stylistic issues, but this hostile acquisition of the em dash by AI and the attendant demand for its jettisoning altogether must be challenged. We simply cannot let AI claim the em dash. It belongs to human writers. It helps the writer emphatically force open a sentence — and it does so with a flair — to provide additional context for the most discerning and perceptive readers.
Even AI detection software designed to detect the content generated using AI is full of blemishes — and prejudices, one must say. Many people have claimed that their assignments — written entirely from scratch after excruciating research — were flagged as AI-generated. The detection software provides false positives because students from formerly colonised countries write English differently from native English speakers. Our English, rooted as it is in a context where everyone puts a premium on an enriched vocabulary, tends to be naturally flowery and sometimes — oftentimes? — ostentatious. Thus, one is always ‘honoured’ for the most paltry of things; a professor, or, for that matter, a speaker, is always ‘esteemed’, and their lecture/session is invariably ‘enlightening’; we always express our ‘sincerest gratitude’. The most mundane of things are laced with grandiloquence. This leads the software to attribute human-written prose to artificial intelligence.
People have always chastised excessively embellished prose, particularly when such embellishment rendered the text turgid and impenetrable. One should have no qualms about this, for the primary purpose of elegant prose is to make the process of reading enjoyable, and if it fails at that, it defeats the very purpose. Scholars like Judith Butler, Gayatri Spivak, Jacques Derrida, etc., have drawn flak from many quarters for the pedantry and elusiveness of their prose, which obscures more than elucidates. So obtrusive is the prolixity — it is deliberate — that Terry Eagleton, a brilliant Marxist literary theorist, argues that even intellectuals can’t quite grapple with what Spivak writes.
However, what we are witnessing now is unprecedented. So pervasive is the paranoia stoked by AI that any hint of elegance, any attempt at theorisation, is seen with suspicion these days, if not altogether castigated as AI-generated. This, I believe, is a natural extension of the anti-intellectualism that was never really dormant in our society. Although it eludes a precise definition, an article published in Studio Atao proffers a working one: “(Anti-intellectualism) is a social attitude that systematically undermines science-based facts, academic and institutional authorities, and the pursuit of theory and knowledge.”
I am a fierce critic of elitism and pedantry, but it must be acknowledged without any hesitation that we have made a fetish of the virtues of ‘accessibility’, thereby reducing it to an empty abstraction. Anti-intellectualism often remains cloaked under the guise of accessibility. Ostentatious jargon-mongering is, of course, undesirable; it is a mechanism to gatekeep knowledge — paradoxically, the very knowledge that purports to be ‘emancipatory’ and addressed to the ‘subaltern’. However, some people are wont to conflate ‘concepts’ with jargon. The utility of the former must be emphasised, and the tendency to reduce it as a convenient shorthand for the latter vehemently discouraged. Learning and engaging with new concepts is a sine qua non for any progressive politics. The populist slander against theorisation is not only an expression of misguided hostility against elitism; it also serves as a substratum to bolster reactionary politics. This faux radicalism serves to torpedo change and encourages an uncritical espousal of ossified ideas — notwithstanding evidence to the contrary.
The writer cannot patronise the reader — it is true — and must be grateful to them, but they are also under no compulsion to constantly pamper the reader with crude simplicity. Furthermore, one could also invert the value judgement and beseech a different question: is it not condescending or patronising to assume that ‘people’ cannot grapple with complexity, and therefore everything should be dumbed down for them? Here, I am reminded of a particularly colourful paragraph from Rosa Luxembourg’s Reform and Revolution that does a lot of significant unpacking:
“No coarser insult, no baser aspersion, can be thrown against the workers than the remarks, “Theoretical controversies are only for the intellectuals.” Lassalle once said, “Only when science and the workers, these opposite poles of society, become one will they crush in their arms of steel all obstacles to culture. The entire strength of the modern labour movement rests on theoretical knowledge.” As long as theoretical knowledge remains the privilege of a handful of intellectuals in the Party, it will face the danger of going astray. Only when the great mass of workers take in their own hands the keen and dependable weapons of scientific socialism will all the petty-bourgeois inclinations, all the opportunistic currents, come to naught.”
Thus, artificial intelligence and anti-intellectualism — two AIs — share a fascinating dialectic. With the burgeoning of AI, anti-intellectualism has found a new and insidious alibi. Instead of critically engaging with a text, people conveniently ascribe it to artificial intelligence.
Under these circumstances, there can’t be one fit-for-all solution. The flourish of chatbots is real; their uncritical deployment by humans to produce generic drivel is real, but it is also true that many people — students, in particular — are unfairly accused, which can potentially jeopardise their academic careers. We must critique the misuse of these tools, but as the popular saying goes, we must be fleet-footed enough not to throw out the baby with the bathwater : we must not capitulate to this culture of paranoia. Above all, we must emphasise the usefulness of the em dash; otherwise, it would be rendered obsolete in no time. Let’s not rush to discard it, for doing so would entail ceding a piece of our literary and intellectual inheritance. We certainly cannot afford that.
Shashi Singh is pursuing a master’s in Delhi University.