In the revelatory closing pages of his 2024 novel Playground, Richard Powers has one of his narrators reflect on the arrival into the world of ChatGPT.
We are at an indeterminate point in the future, and the technology is unnamed. But we recognise it from the familiar story that its “overnight appearance rocked the world and divided humanity”.
“Some people saw glimmers of real understanding. Others saw only a pathetic pattern-completer committing all kinds of silly errors even a child wouldn’t make.”
It turns out, however, that this version of what we’ve learned to call a Large Language Model (LLM) was only the beginning.
‘All is being done’ to secure release of Irish woman kidnapped from Haiti orphanage, says Tánaiste
Bundee Aki reveals wife gave birth in a car hours before he played for Lions in first Test against Australia
‘If we keep knocking down places like Smyth’s, soon there’ll be nothing interesting left for tourists to visit’
Laura Kennedy: Body positivity fell from favour as soon as Ozempic brought thinness back
Later improvements multiply the technology’s power beyond imagination. Playground describes these developments with a tone of inevitability, as a question of when, rather than if.
In a profile of him in the Spanish newspaper El País, I was struck by a single blunt statement by Powers: “AI represents the complete victory of capital over labour”.
Here was a sentiment – registered in a time-honoured Marxist vocabulary – that seemed absent from Playground, or at least significantly muted within it. The technology is mostly seen in the novel through the sympathetically rendered consciousness of a Big Tech magnate.
Another strand considers AI’s implications for the inhabitants of a small Pacific island. But its broader effects on the lives of workers or the market for jobs remain relatively unexplored.
[ Tech shocks to industry have only just begunOpens in new window ]
This division between the prophetic speculations of Playground and the hard-nosed political analysis of its author is a version of the “divided humanity” to which the novel alludes.
Just as there are those who see “glimmers of real understanding” in LLMs and are excited by the potential for untold technological breakthrough, there are others who view this “pathetic pattern-completer” as robbing humans of the kind of cognitive labour that, on some accounts, is the core of who we are.
Somewhere in the middle are the many – perhaps the overwhelming majority – for whom ChatGPT is just one more technology to adapt to, to use where one can, with the larger consequences of its widespread adoption lying somewhere out of sight.
These thoughts have practical application to my day job as a lecturer in English (can you tell?) at University College Dublin.
As in universities the world over, the last two years have seen an unprecedented challenge by LLM technology to our assessment procedures, which have until now centred on the analytical essay. Over recent decades, for a variety of good reasons, we have moved away from in-person exams towards more varied forms of assessment, but the take-home essay has remained at the heart of our pedagogy. All this now looks set to change.
Unless you know a student well, marking an essay has become a battle between blind faith and cynicism. One of the most insidious effects of LLMs is that even good student work comes under suspicion as potentially the product of a machine.
As the technology improves, the temptation to use it to game the system grows. This represents – to adopt Marxist language again – a form of alienation: the language we use is no longer our own.
[ We need to talk about AI’s staggering ecological impactOpens in new window ]
And this is before we even reckon with the vast amounts of electricity and fresh water needed to power this technology, along with the exploited workers in the Global South that, as Karen Hao reports in Empire of AI, are required to improve its outcomes and clean up its language.
Powers is right that AI represents the complete victory of capital over labour. That victory has been so fast and so discombobulating that it is hard to know on what ground to stand, never mind how to fight back. To consider AI part of a ruling class project is not to imagine it as a conspiracy, but simply to acknowledge that its effects align with the aims of the powerful. For all the talk about boosting productivity, authoritative studies have very quickly demonstrated that LLM use makes people less willing and able to think.
The most widely reported study of this kind, conducted by the Media Lab at MIT, found lower levels of brain engagement among participants asked to use ChatGPT to write essays, by comparison with those writing them without such support.
Other studies, this time in political science rather than neuropsychology, demonstrate that voting Republican in US elections is strongly correlated with not having a university degree.
Since taking power, the Trump administration has launched an unprecedented war on universities, attacking not only humanities and social science programmes but also the previously untouchable architecture of STEM research. Top academics are leaving the US in droves. There is talk in Ireland and elsewhere of benefiting from this brain drain.
[ ‘Really scary territory’: AI’s increasing role in undermining democracyOpens in new window ]
At the same time, Trump and his cronies are pumping billions into AI research. This is not a coincidence. LLMs are the most effective tool ever created to curtail the traditional work of universities, the cultivation of critical individual minds. In these circumstances even those with university degrees, the pattern suggests, will become more likely to vote for Trump and his ilk. This in turn will enable further legislative abominations like the “big beautiful bill,” which deprives the poorest of healthcare while lining the pockets of the already astronomically rich.
In universities, we worry not only about student writing but about student reading. Anecdotal evidence suggests that students are becoming less willing and able to read the kinds of long novels that used to be the core of an English degree.
This is not only a question of technology but of political economy, with many students now having to work part-time or even full time to support their studies. My colleagues and I have responded by gradually reducing the number of set texts on our courses. The sense of inevitability is hard to gainsay.
I would love to have my students read and debate Playground, a novel that speaks to its moment like few artworks do. Is it too credulous about AI? Too cynical? Does it capture the world as they see it? Does it transform their perception of that world?
But setting a multilayered and challenging text for students to read carries newfound risks. Asking them to write about such a text carries even more.
Whirring away in the background, powered by looming data centres, ChatGPT stands ready to do a serviceable job in our absence.
Dr Adam Kelly is associate professor in the School of English, Drama and Film at University College Dublin
