Teaching
Tomorrow: Can AI Rebuild the Curriculum from Scratch?
The school curriculum was not designed for the world that exists
today. It was shaped by industrial-era priorities, refined during the
twentieth century, and inherited largely intact by systems that now attempt
to prepare students for careers, challenges, and technologies that did not
exist when the framework was written. Artificial intelligence has arrived as
both a disruption and an opportunity, forcing educators, policymakers, and
technologists to ask a question that has no comfortable answer: should AI
help us fix what we have, or help us build something entirely
new?
The pressure is real. Students entering primary school today will
enter the workforce in the early 2030s, competing in labour markets already
being reshaped by automation, operating in societies where algorithmic
systems make consequential decisions about credit, healthcare, and
employment. Whether their education prepares them for that reality depends on
choices being made right now, in school districts and ministries of education
worldwide.
Adaptive Learning at Scale
The most immediate application of AI in education is personalised
learning. Platforms powered by machine learning monitor individual student
performance in real time, identify conceptual gaps, adjust the difficulty and
format of exercises, and flag students who are falling behind before they
fall too far. The ambition is an education system that responds to each
learner rather than teaching to an average that fits almost nobody
perfectly.
The US
Department of Education’s 2024 AI guidance report positions
adaptive learning as a central opportunity, arguing that AI-enabled
personalisation can address persistent equity gaps in educational outcomes.
Students with learning disabilities, language barriers, or interrupted
schooling histories stand to gain disproportionately from systems that meet
them where they are rather than demanding they keep pace with a class
cohort.
This is the optimistic case, and there is evidence to support it.
Platforms deploying adaptive algorithms have demonstrated measurable
improvements in mathematics attainment in controlled pilots, particularly for
students in lower-income school districts where teacher-to-student ratios
make individual attention difficult. The technology, at its best, is a force
multiplier for educators working in under-resourced
environments.
The Teacher’s Evolving Role
Personalised AI platforms do not replace teachers. They reframe
what teaching means. When an algorithm handles diagnostic assessment and
practice exercise differentiation, the educator is freed to focus on what
machines handle poorly: mentorship, motivation, ethical reasoning, creative
challenge, and the social dynamics of learning in community.
Research from Science
News Today indicates that teachers who integrate AI tools
effectively tend to shift toward facilitation roles, guiding inquiry and
project-based learning rather than delivering content. This transition is not
automatic. It requires professional development that most school systems have
not yet invested in at scale.
A RAND
Corporation study found that while a substantial majority of US
teachers had experimented with AI-based classroom tools, fewer than a third
had received any formal training on their use. The gap between tool
availability and professional capability is one of the central challenges
facing AI-enhanced education. Technology deployed without training can be as
harmful as no technology at all, particularly when automated systems make
consequential decisions about students’ learning
trajectories.
What the Curriculum Misses
Perhaps the deepest question AI raises about education is not how
to teach but what to teach. As automation displaces routine cognitive and
manual tasks, the economic premium on skills that machines handle poorly is
rising. Critical thinking, creative problem-solving, ethical reasoning,
intercultural communication, and emotional intelligence are consistently
cited by employers and economists as the competencies most resistant to
algorithmic substitution.
Yet these are precisely the subjects least amenable to
standardised assessment, which drives curriculum prioritisation in most national
systems. It is easier to measure whether a student has memorised a historical
date than whether they can construct a nuanced argument about its
significance. AI can help with the former. Only thoughtfully designed human
instruction develops the latter.
As AI
in Theatre explored, the arts and humanities are not immune to
algorithmic assistance, but they remain domains where human judgment,
creativity, and empathy are irreplaceable at the highest levels. Educational
systems that allow these subjects to be crowded out by STEM priorities,
driven by a narrow reading of AI’s economic impact, may be producing graduates
ill-equipped for the human dimensions of working alongside
AI.
Bias in the Classroom Algorithm
The same concerns that arise when AI is applied in hiring,
criminal justice, and healthcare apply with equal force in education.
Algorithms trained on historical data inherit the inequities embedded in that
data. A model trained predominantly on learning outcomes from high-income,
majority-group student populations may perform poorly for students whose
educational trajectories look different from the training
set.
The Learning
Policy Institute has documented how data-driven educational tools
can systematically underperform for students from lower socioeconomic
backgrounds, students of colour, English language learners, and students with
disabilities. When a personalised learning platform recommends a
lower-complexity curriculum pathway based on flawed initial assessment, it
risks locking students into tracks that limit rather than expand their
opportunities.
This mirrors concerns raised across AI applications in contact
with vulnerable populations. As AI
Refugee Forecasting noted, predictive systems operating on marginalised
communities require exceptional care to avoid encoding and amplifying
existing disadvantages.
Rebuilding or Retrofitting?
The most ambitious educators and policymakers argue that AI’s
arrival should prompt not incremental adjustment but fundamental redesign.
The Center
for Humane Technology advocates reimagining schooling around the
capabilities that humans possess and machines do not, rather than continuing
to teach competencies that automation is progressively
eroding.
That means curricula built around ethical reasoning about
technology, collaborative problem-solving, project-based learning that mimics
genuine professional challenge, and explicit instruction in the critical
evaluation of AI-generated information. It means assessing whether students
can navigate ambiguity rather than reproduce memorised answers. It means
preparing young people not just to work alongside AI systems but to question,
audit, and hold those systems accountable.
This is a different kind of education from the one most systems currently
deliver. Building it will require not just AI tools but political will,
sustained investment in teacher development, and genuine engagement with the
communities that schools serve. AI can contribute to that work. It cannot
substitute for it.
The Blueprint Must Remain Human
AI is already reshaping education in ways that are both promising
and perilous. Its capacity to personalise at scale, identify struggling
learners early, and reduce the administrative burden on teachers represents
genuine value. Its potential to embed bias, erode equity, and reduce
education to optimised skill delivery represents genuine
risk.
The curriculum of tomorrow cannot be written by algorithms alone.
It must be co-authored by educators who understand both the possibilities and
the limits of the technology, by communities who know what they need their
schools to produce, and by students who will inhabit the future that
curriculum is meant to prepare them for. AI can scaffold the work. The
blueprint must remain human.
The stakes extend beyond classrooms. A generation educated to
think critically about AI, to understand its capabilities and its limits, to
interrogate the data it is trained on and the values it embeds, is a
generation better equipped to participate meaningfully in the governance of
these systems. As explored in The
Forgotten Accent, the communities most affected by AI’s cultural
assumptions are often the least represented in the rooms where those systems
are designed. Education is one of the few levers capable of changing that
over time. Getting it right matters more than getting it
fast.
About the Author
Stuart Kerr is the Technology Correspondent for LiveAIWire,
covering artificial intelligence, ethics, and the ways technology is
reshaping everyday life.