The general purpose language model — as exhibited in programs like ChatGPT that interpret and produce writing automatically — is a fundamentally liberatory technology that has the potential to free working people from the busywork of mechanical writing. Unless something is done, we will not be allowed to use it as such.

The recent controversy against ChatGPT has had two distinct fronts. The first critique is one of skepticism. On the one hand, ChatGPT is unsophisticated: it misinterprets complex prompts and writes in a manner that is clearly technical with no style imparted. Language models lack a uniquely human creativity. The other critique is one of fear. Critics are concerned with the general purpose language model’s existential ability to erase writing as a human skill wholecloth. Students will turn in undetectable AI-generated essays. Our great novelists will be outmoded by AI-generated best-sellers. Writing will become obsolete as humanity is replaced by its gizmos.

As these arguments are contradictory, I’ll choose to analyze the first one; ChatGPT is best at writing mechanically, which means that it is ideal for automating menial forms of writing. Language models are best at replicating sources with many examples that conform to a certain template or pattern. In other words, ChatGPT cannot write “Infinite Jest,” but it is pretty good at writing genre fiction and nearly flawless at writing professional emails or neutral bullet-point summary news articles. Language models do not replace innovation — they replace busywork and menial writing that forces writers to sacrifice their time and autonomy to mechanical, socially rigid formats. ChatGPT should be understood not as an author but as a translator of common language prompts into the syntax of professional memos, five-paragraph essays and programming languages.

As it currently exists — freely accessible to anyone with an internet connection — this translatory function of ChatGPT has the potential to flatten epistemic inequalities driven by class disparity. Translational gaps have been utilized to perpetuate social hierarchy throughout human history. Take English, for example. Just as French had been the official language of English government officials in the 11th-14th centuries, the British empire enforced English onto its colonial subjects until English became the de facto official language of global trade. Even within the English language, African American Vernacular English has been historically discriminated against by the white ruling class as unprofessional, resulting in racist hiring practices and the development of social techniques like code-switching. In each case, linguistic formats made unevenly available due to socioeconomic inequality have stood as barriers to power. Translation technology closes these gaps by breaking through these epistemic barriers, and ChatGPT is no exception.

 Instead of simply translating between languages, ChatGPT translates between modes of writing within a language: prompt to essay, essay to summary, bullet-point notes to email and so on. What is immediately unique about ChatGPT is its ability to translate large aggregated datasets into comprehensible explanations and its consequent ability to translate natural language into programming language. At the same moment that entry-level coding ability has become an essential requirement for many living-wage jobs, another epistemic barrier to power, ChatGPT now allows all literate people the ability to produce basic, functional code.

 There is a great egalitarian promise here: all of human knowledge at our fingertips, translated to be comprehensible and workable. I feel a dreadful urgency to compel you to see what you can do with this technology while it is new and volatile and publicly accessible. Have you ever wanted to build a website? Is there some task at your job that feels like it could be automated? Would you like to never write a professional email again? All of these things are now possible for virtually everyone. Do them! Then, ask yourself what else you can do in all the time you have saved. Let a hundred flowers bloom! Let a thousand machines automate our labor so that we might take our free time to bathe in the sun and blossom as humans. Take advantage of the language model before it is incorporated into the economic framework of our society, as it inevitably will be.

This technology is revolutionary in the same sense that the textile mill or steam engine were revolutionary, but any economic revolution entails the reorganization of human capital. Just as the Luddites lost employment to the textile mill, journalists are under direct threat from this technology, such as 12% of BuzzFeed’s staff laid-off in late January. Amid contemporary layoffs of journalists and tech workers, the market will come to require different skill sets for writers. Creating complex programs still requires software engineers with the ability to read and edit code in the same way that creating a whole newspaper still requires editors that will fact-check and curate the print. The language model will become part of an iterative cycle of text production that values creativity and curation. 

Yet, as the global economy reorganizes around automated language processing, it will squeeze workers for all they can produce. Americans have been victim to a growing gap between productivity and pay since the 1980s. From 1950-2019, while GDP per capita increased by 293%, mean working hours only decreased by 11%. When you tell your boss that you completed your coding quota for the day in an hour using ChatGPT, he will not tell you to go home, nor will he pay you more. He will quadruple your quota and half your staff, and if he fails to do so then he will be beaten by his competitors that do. No matter how much we automate, we will labor as long as economic growth directs our society.

The liberatory potential of ChatGPT — the automation of machinic writing in favor of increased free-time — will remain unrealized due to the global economy’s demand for growth. Automation does not allow us to work less because the market requires firms to produce more. Still, the broad accessibility of language models has placed previously unimagined informational and productive power into the hands of everyday people. If the corporations cannot use this technology to improve how we live, then it is up to the people. 

Alexei Gannon, FCRH ’25, is a biology and history major from Allentown, Penn.