BEYOND
THE STOCHASTIC PARROT: RECLAIMING HUMANITIES IN THE AGE OF AI
KALYANI JAIN
Jindal School of International Affairs, O.P.
Jindal Global University.
ABSTRACT
The rapid proliferation
of Generative Artificial Intelligence (AI) has triggered an existential crisis
within higher education, as algorithms demonstrate the ability to synthesize
historical data and draft complex analytical essays. However, recent empirical
labour market data presents a profound puzzle: despite high indices of AI
occupational exposure, university professors are experiencing job expansion
that outpaces decline. Drawing on the World Economic Forum’s ‘Future of
Jobs Report 2025’ and the International Labour Organization’s ‘Working
Paper 140’ (2025), this paper investigates this ‘Professor Paradox.’ It
explores why the academic profession at university level broadly resists
automation. By contrasting it with highly vulnerable administrative roles, it highlights
the critical macroeconomic distinction between task-level augmentation and
occupational replacement. To contextualise this technological disruption, the
paper conducts a historical analysis of the Indian education system, which
remains deeply entrenched in the 19th century colonial Macaulay
model of rote, clerical learning. The paper argues that while the education
system has not yet moved away from this output-based paradigm, the AI
revolution provides an urgent, existential catalyst to finally dismantle it.
Generative AI excels at syntactic replication and data retrieval but
fundamentally lacks the capacity for ethical reasoning, historical empathy, and
contextual meaning-making. By relying on algorithmic shortcuts to bypass the
friction of critical thought, future generations risk losing the very
intellectual and empathetic capacities that define human citizenship.
Ultimately, the paper concludes that the Humanities are not rendered obsolete
by AI. Instead, by forcing a necessary departure from colonial pedagogies
toward true dialectical friction, AI elevates the Humanities to the highest
imperative of democratic defence.
Keywords: Artificial
Intelligence, Labour Economics, Humanities, Macaulayism.
INTRODUCTION
In the wake of the
global deployment of Large Language Models (LLMs) such as ChatGPT, Claude, and
Gemini, a palpable sense of existential dread permeated the halls of academia.
For centuries, disciplines such as History, Sociology, and Political Science have
assessed student competence through the written word, such as analytical
essays, historical research papers, and policy briefs. When Generative AI
demonstrated the ability to produce these exact artifacts in seconds, often
with a syntactic fluency rivalling undergraduate levels, it seemed to herald
the obsolescence of the Social Scientist. If an algorithm can synthesise
historical timelines, pass standardised competitive examinations, or draft a
Political Science literature review, what is the future economic and societal
utility of the Humanities?
The answer to this
question does not seem to be as straightforward as one would expect. A
macroscopic view of current labour trends reveals a glaring empirical puzzle.
According to the World Economic Forum’s ‘Future of Jobs Report 2025’,
specifically analysing job growth versus decline by occupation, university and
higher education teachers show a projected scope of expansion that
significantly outpaces their projected decline. This paradoxical trend is robustly
corroborated by the International Labour Organization (ILO 2025), whose
empirical analyses of Generative AI's impact on employment indicate that
educators face massive “exposure” to AI but minimal actual job destruction.
How can an occupation
be simultaneously highly exposed to advanced automation and still be rapidly
expanding?
This paper seeks to
resolve this ‘Professor Paradox’ by rigorously examining both the macroeconomic
mechanics of labour and the historical purpose of the Humanities. If the
function of a university was merely to transfer information and produce
standardised written outputs, Generative AI would indeed render human educators
obsolete. However, as Türkkahraman (2012) and Murray (2023) emphasise,
education serves as the fundamental engine for societal development, which is
not just about data retrieval, but the act of civic cultivation, ethical
moderation, and the development of democratic judgment.
To resolve this
paradox and understand why the academic profession is insulated from the very
technology that mimics its outputs, this paper breaks down the macroeconomic
mechanisms at play. It contrasts academic expansion with the specific
administrative jobs that are being destroyed by AI (such as data
entry and telemarketing) to precisely isolate the ‘human elements’ that
algorithms cannot replicate. Furthermore, it situates this technological crisis
within the history of Indian education. The current academic system is still
largely trapped in the 1835 Macaulay framework of clerical, rote learning. This
paper argues that AI is not destroying human intellect; rather, it is
destroying the clerical model of education. Ultimately, the AI revolution
presents an urgent, unavoidable opportunity to abandon our colonial educational
hangover and reclaim the uniquely human elements of historical empathy and
ethical reasoning, lest the coming generations lose their humanity entirely to
cognitive offloading.
THEORETICAL FRAMEWORK
To move beyond a
purely descriptive analysis of technological disruption, this paper utilizes
three intersecting theoretical frameworks drawn from labour economics,
political science, and cognitive psychology.
First, the paper
adopts the Automation of Work Framework proposed by Melián-González
and Bulchand-Gidumal (2025). Moving away from traditional, flawed
‘job-cantered’ approaches that falsely predict total occupational replacement,
this framework requires multiple scopes of analysis, specifically breaking down
labour into distinct duties and tasks. Applying this framework reveals that the
complete displacement of complex roles is highly improbable. Instead, AI
drives partial job automation and work redesign. Generative AI automates
the routine clerical tasks of academia, forcing a structural redesign of the
educator's role toward non-routine, interpersonal, and empathetic duties that
technology cannot replicate.
Second, the paper
utilizes Historical Institutionalism to analyse the persistent legacy
of British colonial education. Historical institutionalism emphasizes the
concept of ‘path dependence’, i.e., the tendency of institutions to remain
stable and highly resistant to change over long periods (Mohn, 2024.). For
nearly two centuries, the Indian education system has remained path-dependent
on Thomas Macaulay’s 1835 clerical model. However, historical institutionalism
also identifies ‘critical junctures’, i.e., pivotal moments of systemic
disruption that present a brief window to radically alter an institution's
trajectory. This framework posits that Generative AI acts as the ultimate
critical juncture. Because AI is a flawless ‘clerk,’ the institution must
finally abandon its path-dependent colonial model or render human cognition
economically worthless.
Third, the paper
synthesises the psychological concept of Cognitive Offloading with
John Dewey’s Theory of Education. ‘Cognitive offloading’ refers to the human
tendency to use external devices to reduce the mental effort required to
perform a task, which, over time, can lead to severe cognitive ‘skill decay’
(Risko and Gilbert, 2016). Recent empirical studies confirm that individuals
increasingly trust and rely on technological tools to manage memory demands,
exacerbating the risk of offloading (Peng and Yeh, 2025). This psychological
risk directly threatens the sociological imperatives outlined by Dewey. In
revisiting Dewey for the modern algorithmic age, Peters and Jandrić (2017)
emphasise that participatory democracy requires the active friction of human
experience. If students permanently offload their analytical thinking to LLMs,
they bypass the productive cognitive friction required to develop moral
judgment.
METHODOLOGY
Because this paper
addresses a large-scale structural shift at the intersection of technology,
labour economics, and educational philosophy, it relies on a mixed-methods
approach utilising secondary economic data analysis and comparative historical
analysis.
To establish the
central puzzle of occupational expansion despite high AI exposure, the paper
utilises the World Economic Forum’s ‘Future of Jobs Report 2025’,
specifically analysing its demographic projections for educational and
administrative sectors. To explain the mechanics of this paradox and isolate
the defining ‘human element,’ the study utilises the International Labour
Organization's ‘Working Paper 140’ (2025). By contrasting the projected growth
of academia and social sciences against the projected collapse of strictly
clerical roles, the methodology quantifies exactly what labour markets value in
human workers in an algorithmic age.
To determine if the
current panic over AI is uniquely dangerous or a repetition of past
technological anxieties, the paper employs historical comparison. It examines
the ancient Socratic fear of the written word (detailed in Plato’s Phaedrus)
as the original ‘cognitive panic.’ It then heavily contrasts the modern
capabilities of Generative AI against the 19th century
implementation of the Macaulay education system in India. By tracing the
historical through-line of ‘clerical education,’ the methodology demonstrates
that AI is shifting the fundamental nature of the social scientist's job,
rather than eliminating the need for the discipline itself.
RESOLVING THE CORE
PUZZLE
To unpack the
‘Professor Paradox,’ one must rigorously examine why certain jobs are entirely
insulated from AI while others are facing imminent collapse. The ILO's 2025
Working Paper provides the definitive economic answer, i.e., high exposure to
generative technology does not equal occupational replacement.
According to the WEF
and ILO reports, the jobs facing rapid, systemic decline, such as bank tellers,
postal clerks, data entry keyers, payroll clerks, and administrative
secretaries, all share a defining characteristic. They are structurally devoid
of the ‘human element.’ These roles require the routine, syntactic processing
of data without the need to assign emotional, ethical, or historical meaning to
that data. Because Generative AI is a flawless execution engine for routine
cognitive tasks, these jobs are rendered obsolete.
Conversely, in the
academic and Social Science sectors, Generative AI acts as a profound augmenter
rather than a replacer. For a university Professor or a policy researcher, AI
primarily targets the routine administrative drudgery of the profession like
synthesising massive literature reviews, translating texts, checking for
plagiarism, or grading standardized multiple-choice assessments. Liberated from
this clerical burden, the professor’s productivity exponentially increases,
allowing a full pedagogical shift to what machines cannot do. This stark
bifurcation of the labour market, between clerical replacement and humanistic
augmentation, is illustrated in Figure 1.
This raises the
critical question: What exactly is the ‘human element’ in Humanities that insulates
it from automation? Based on the limitations of LLMs, this element consists of
three un-automatable capacities:
Furthermore, there is
a macro-structural reason for academic expansion. As AI displaces millions of
routine clerical workers across the globe, the State and private enterprise are
being forced to launch massive retraining initiatives. According to the macro-economic
projections of the World Economic Forum (2025), the global labour market is
undergoing an unprecedented structural churn centred around reskilling and
upskilling. Displaced workers are returning to universities, vocational
schools, and public policy institutes to learn AI-resilient, human-centric
skills. This massive influx of adult learners is directly driving the projected
job growth for educators and social scientists. Thus, the paradox is resolved:
AI destroys clerical work, which in turn drives massive demand for humanistic,
ethical education.
THE MACAULAY HANGOVER
While the
institutional role of the professor and the social scientist seems to be
economically secure, the intellectual development of the student is
under an unprecedented historical threat. To truly understand the danger of
Generative AI, one must acknowledge a difficult and uncomfortable truth, i.e.,
the modern Indian education system has not yet moved away from its colonial
origins. We are still operating deeply within the pedagogical framework of
Thomas Babington Macaulay’s ‘Minute on Indian Education’ of 1835.
The explicit
political goal of the British colonial state was not to cultivate independent
philosophers, indigenous scholars, or democratic citizens capable of critical dissent.
Rather, the goal was to mass-produce a clerical class, or “a class of
persons, Indian in blood and colour, but English in taste, in opinions, in
morals, and in intellect” (Macaulay, 1835). These individuals were designed
to serve as administrative intermediaries who could process data and execute
the orders of the colonial state with mechanical precision. To achieve this,
the colonial administration instituted an education system prioritising rote
memorization, the formulaic reproduction of data in written examinations, and
strict syntactic compliance (Viswanathan, 1989).
Nearly two centuries
later, despite the end of colonial rule, our universities still largely reward
Macaulay’s model of learning. We test students on their ability to act like
data-retrieval machines. Success in the modern educational system is still
largely determined by a student's ability to memorize vast quantities of
information and reproduce it in highly standardised, output-based written
examinations.
Generative AI exposes
the absolute intellectual bankruptcy of this ongoing system. ChatGPT is, by
definition, the ultimate colonial clerk. It can memorize historical dates,
retrieve jurisprudence, and draft formulaic, grammatically perfect essays
exponentially faster and more accurately than any human student. If we continue
to operate on this historical model by testing students on their standardised
output rather than their frictional thought process, the education system will
be entirely hollowed out.
Therefore, it can be
argued that AI is not destroying human intellect. Rather, it provides an
urgent, existential catalyst to finally dismantle the Macaulay model. We have
not yet moved away from it, but AI means we now must, lest the upcoming
generations lose their very humanity (as summarised in Figure 2).
If students continue
to use algorithms to bypass the friction of writing and analysing, they risk
experiencing a devastating atrophy of thought. This echoes the most profound
historical parallel to our current crisis, found in classical antiquity. In
Plato’s Phaedrus, Socrates recounts the Egyptian myth of the God
Theuth, who presents the invention of writing to King Thamus. Thamus rejects
the technology, warning that delegating human memory to external tools provides
“not truth, but only the semblance of truth; they [humans] will be hearers of
many things and will have learned nothing; they will appear to be omniscient
and will generally know nothing” (Plato, 360 BC). Today, if a student uses an
LLM to generate an essay on the causes of democratic backsliding, they have
acquired the appearance of wisdom without the underlying intellectual struggle
required to attain true comprehension.
THE HUMANITIES
IMPERATIVE
To understand why
this cognitive atrophy is so dangerous, we must look at how algorithms
function. Computer scientists explicitly warn against treating LLMs as arbiters
of truth, logic, or meaning. Bender et al. (2021) famously categorized Large
Language Models as “stochastic parrots” i.e., systems that blindly stitch together
linguistic sequences based on probabilistic training data.
Generative AI
recognizes patterns of words, but it has no semantic understanding of the real
world. It does not know what a ‘democracy’ is, nor does it understand ‘human
rights.’ It only knows that those words frequently appear near each other in
its training data. Because an algorithm has never possessed a body, faced
mortality, experienced poverty, or felt political oppression, it is
fundamentally disconnected from the human condition. Therefore, when an AI
writes an essay about the Partition of India, the French Revolution, or the
drafting of the Constitution, it is merely calculating statistics, without
being affected by it emotionally.
If we allow the
coming generation to outsource their historical, sociological, and political
analysis to stochastic parrots, we will actively cultivate a society incapable
of empathy and vulnerable to manipulation. For Political S not
truth, but only the semblance of truth; they will be hearers of many things and
will have learned nothing; they will appear to be omniscient and will generally
know nothing; they will be tiresome company, having the show of wisdom without
the reality. Science scholars, this is a matter of democratic survival.
Democracies are inherently fragile ecosystems that rely entirely on the active,
critical participation of their citizens. John Dewey (1916) argued that a
functioning democratic State requires a populace capable of navigating
epistemic uncertainty, recognising propaganda, and engaging in empathetic moral
reasoning with opposing viewpoints. Democracy is not a machine that runs on its
own but is a human system that requires constant human judgment.
Generative AI
environments, often operating as proprietary ‘black boxes’ controlled by
monopolistic technology corporations, are fundamentally anti-democratic in
their intellectual architecture. They are designed to provide singular,
authoritative-sounding answers, artificially flattening the complex, multipolar
debates that define Political science and History. They remove the nuance, the
disagreement, and the friction that is essential to the democratic process.
Furthermore, the
digital public sphere is rapidly being flooded with AI-generated synthetic
media, deepfakes, and automated disinformation campaigns. In this new digital
ecology, technical and clerical skills will not save democratic institutions.
The only defence against a post-truth, algorithmically manipulated society is a
citizenry deeply rooted in the analytical frameworks of the Humanities. The
future citizen must possess the historical literacy to recognise authoritarian
rhetoric, the philosophical grounding to debate ethics, and the sociological
training to spot structural bias in an AI’s output.
To achieve this, the
Social Sciences must drastically and immediately overhaul their pedagogical
approach. If an assignment can be completed flawlessly by an AI in ten seconds,
it is no longer a valid assessment of human intellect. The field must move away
from ‘output-based’ written assessments and return to ‘process-based,’ highly
interactive evaluations. We must embrace Socratic dialogue, oral defences, live
civic debate, and experiential learning. An algorithm cannot participate in the
spontaneous, face-to-face friction of a human debate. It cannot feel the
ethical weight of a policy decision in a classroom, nor can it demonstrate
contextual empathy to a peer.
CONCLUSION
The panic surrounding
the obsolescence of the university professor and the social scientist in the
face of Artificial Intelligence is historically and economically misplaced. As
the empirical data from the WEF and ILO demonstrates, the academic profession
is expanding because human educators are augmented by AI, insulated by
institutional credentialing, and desperately needed to cultivate the
non-routine, deeply human skills required in a disrupted 21st
century workforce.
The true, existential
crisis we face is the potential automation of the student's mind. For two
centuries, we have remained trapped in a colonial education system designed to
produce administrative clerks. Generative AI has now mastered clerical work. We
can no longer afford to test students on their ability to act like algorithms.
Generative AI offers a frictionless, intellectually hollow shortcut that
threatens to atrophy the critical reasoning, ethical judgment, and historical
empathy of the coming generations.
Because democratic
governance is entirely dependent upon an engaged and critically capable
citizenry, the outsourcing of human thought to the algorithmic ‘mind’ poses a
structural, fatal threat to the fabric of free societies. The future does not
seem to render the Humanities obsolete, rather, by forcing us to finally
abandon the Macaulay model and prioritise human friction, meaning making, and
ethical debate, the AI revolution elevates the Social Sciences to the absolute
highest imperative of public policy and democratic defence.
REFERENCES
Bender, E. M., Gebru,
T., McMillan-Major, A., & Shmitchell, S. (2021). On the Dangers of
Stochastic Parrots: Can Language Models Be Too Big? Proceedings of the
2021 ACM Conference on Fairness, Accountability, and Transparency, 610–623.
https://s10251.pcdn.co/pdf/2021-bender-parrots.pdf.
Dewey, J.
(1916). Democracy and Education: An Introduction to the Philosophy of
Education. Macmillan.
International Labour
Organization (ILO). (2025, May). Generative AI and Jobs: A global
analysis of potential effects on job quantity and quality. ILO Working
Paper 140. https://doi.org/10.54394/HETP0387.
Macaulay, T. B.
(1835). Minute on English Education, 2nd February 1835, in H. Sharp (ed.),
Selections from Educational Records Part I, 1781–1839 (Calcutta: Superintendent
Government Printing, 1920), 107–117. In Colonial Education and India
1781-1945 (1st ed., pp. 55–64). Routledge.
Melián-González, S.,
& Bulchand-Gidumal, J. (2025). A framework for research on the automation
of work. Technological Forecasting and Social Change, 215,
124093. https://doi.org/10.1016/j.techfore.2025.124093
Mohn, E. (2024).
Historical institutionalism. Research Starters: Social Sciences and
Humanities. EBSCO. https://www.ebsco.com/research-starters/social-sciences-and-humanities/historical-institutionalism
Murray, J. (2023).
What is the purpose of education? A context for early childhood
education. International Journal of Early Years Education, 31(3),
571 578. https://doi.org/10.1080/09669760.2023.2238399
Peng, J.-L., &
Yeh, S.-L. (2025). Cognitive Offloading in Short-Term Memory Tasks: Trust
Toward Tools as a Moderator. International Journal of Human–Computer
Interaction, 41(21), 13382-13393. https://doi.org/10.1080/10447318.2025.2474449
Peters, M. A., & Jandrić, P. (2017). Dewey's Democracy and Education in the age of digital reason: the global, ecological and digital turns. Open Review of Educational Research, 4(1), 205–218. https://doi.org/10.1080/23265507.2017.1395290
Plato (370 BCE.) Phaedrus. Acadia University. https://socrates.acadiau.ca/courses/engl/rcunningham/1103/Phaedrus.pdf.
Risko, E. F., &
Gilbert, S. J. (2016). Cognitive offloading. Trends in Cognitive
Sciences, 20(9), 676-688. https://doi.org/10.1016/j.tics.2016.07.002
Türkkahraman, M.
(2012). The role of education in the societal development. Journal of
Educational and Instructional Studies in the World, 2(4), 38-41. https://www.jstor.org/stable/43155369
Viswanathan, G.
(1989). Masks of Conquest: Literary Study and British Rule in India.
Columbia University Press.
World Economic Forum.
(2025, January). Future of Jobs Report 2025. https://www.weforum.org/reports/the-future-of-jobs-report-2025/.