Artificial intelligence is often described as a revolutionary force in education, but history suggests it may be the latest chapter in a much longer story of technology reshaping how we teach, learn and assess.
This blog explores AI’s place in that lineage, and what it means for those designing and delivering assessment today.
From the printing press to teaching machines to AI
Education has always evolved alongside technology.
- The Printing Press in the 15th century radically expanded access to knowledge, moving learning beyond clergy and elites and enabling independent study through textbooks.
- Blackboards, then whiteboards and digital displays, unified the classroom experience and paved the way for modern multimedia teaching.
- The Monitorial System in the 19th century scaled teaching by having more advanced pupils teach their peers, following scripted methods – an early attempt to “industrialise” education.
- Skinner’s Teaching Machines in the 1950s were a major step towards personalised learning. These devices provided immediate feedback, only moving learners forward when they answered correctly, and were grounded in behaviourist principles and mastery learning.
Today, AI-powered tools, from adaptive tutors to guided learning platforms, echo many of these same ideas: breaking learning into small steps, offering real-time feedback and freeing teachers from some forms of routine delivery. The difference is that modern AI can interpret much more nuanced input and adjust dynamically, rather than following a fixed script.
Assessment at the centre of change
Assessment sits at the intersection of pedagogy, technology and society. It does far more than measure attainment:
- Assessment for Learning (AfL) helps teachers and students understand where learning is secure and where support is needed, reducing the risk of “teaching in the dark”.
- Assessment of Learning (AoL) underpins standards, progression and professional accountability.
- When well designed, assessment can support social mobility by fairly recognising the capabilities of learners from diverse backgrounds. When poorly designed, it can reinforce existing inequalities through cultural, linguistic or contextual bias.
Over the past century, assessment has been heavily influenced by what might be called the 'age of efficiency and testing' multiple-choice tests, IQ tests and tightly specified outcomes aligned with Taylorist, factory-style thinking. This has brought scalability and standardisation, but also concerns about narrowed curricula, test anxiety and a focus on exam technique over deeper learning.
AI now arrives into this complex environment, not as a neutral tool, but as another force that can either challenge or reinforce existing paradigms.
Using the SAMR model to understand EdTech and AI
The SAMR model (Substitution, Augmentation, Modification, Redefinition) offers a helpful lens for understanding how technology changes practice:
- Substitution – Technology replaces an existing method with little functional change.
Example: Moving paper exams to on-screen tests to improve logistics, while the core assessment model stays the same. - Augmentation – Technology improves efficiency or quality but within current paradigms.
Example: Using AI to provide real-time feedback on writing, or to support teachers with marking and planning, shortening feedback loops and making formative assessment more powerful. - Modification – Technology enables significant redesign of tasks and processes.
Example: Micro-credentials and blockchain-backed evidence portfolios that build a cumulative picture of competence rather than relying on a single, high-stakes exam. - Redefinition – Technology makes entirely new forms of assessment possible.
Example: Neuro-assessment and immersive XR/VR simulations that assess how a learner thinks, manages cognitive load and responds emotionally in realistic scenarios, rather than only what they can write in an exam booklet.
Skinner’s Teaching Machines arguably sat in the Augmentation category, strengthening existing approaches to teaching and assessment. In contrast, AI has the potential to push assessment towards Modification and even Redefinition, especially when combined with immersive and biometric technologies.
AI through the lens of Gartner’s Hype Cycle
The Gartner Hype Cycle helps explain how expectations around new technologies typically unfold:
- Innovation Trigger
- Peak of Inflated Expectations
- Trough of Disillusionment
- Slope of Enlightenment
- Plateau of Productivity
AI in education appears to be moving from the Peak of Inflated Expectations towards the Trough of Disillusionment. Early headlines about “AI replacing teachers” or “ending marking” are giving way to more sober conversations about:
- Hallucinations and bias in AI outputs
- Risks to academic integrity
- The reality that many learners use AI as a shortcut rather than a learning partner
- The workload involved in safely and effectively integrating AI into teaching and assessment (the “AI paradox”)
There is no guarantee that AI will reach the Plateau of Productivity in education. Economic, environmental and ethical challenges – from the cost and energy demands of AI infrastructure to concerns over “AI-generated workslop” – must be addressed.
However, history suggests that, with time, technologies often settle into more realistic, sustainable roles than either the hype or the backlash initially predict.
What this means for policy makers and practitioners
Looking at AI through historical and conceptual lenses leads to several practical implications:
- Learn from past transitions (e.g. calculators, online learning, digital assessment). These did not destroy core educational goals but did require thoughtful curriculum and assessment redesign.
- Focus on alignment, not resistance. Rather than trying to make assessment “AI-proof”, aim to design tasks that remain meaningful in a world where AI is normal – emphasising reasoning, application, collaboration and authenticity.
- Balance efficiency with humanity. AI can support faster feedback, richer data and more personalised experiences. But a narrow focus on efficiency risks further dehumanising education and eroding the creative and relational aspects that matter most.
- Use frameworks deliberately. SAMR and the Hype Cycle will not predict the future, but they provide structured ways to evaluate new use cases and to separate sustainable innovation from short-lived hype.
- Keep equity at the centre. Access to AI, digital infrastructure and high-quality support will vary widely. Policy and design decisions must aim to reduce, not widen, existing gaps in opportunity.
AI as continuation, and possible turning point
In many ways, AI is a continuation of a long lineage of education technology: from the Printing Press to teaching machines, from broadcast media to online platforms. Each wave has changed what is possible, while also revealing the limits of purely technological solutions.
What is different now is AI’s generative, adaptive nature. Instead of simply providing content, it can co-create explanations, feedback and assessments with learners and teachers. Information is no longer just searchable; it can be dynamically synthesised and contextualised.
Whether this ultimately amounts to a true redefinition of teaching, learning and assessment will depend less on the technology itself and more on how policy makers, leaders and practitioners choose to integrate it into the fabric of education.
By grounding our decisions in historical insight, robust pedagogy and a clear-eyed view of AI’s strengths and weaknesses, we can move beyond hype – and work towards an assessment system that is fairer, more responsive and more genuinely aligned with the needs of learners in the 21st century.
Do you want to be at the the forefront of transforming assessment?
Discover what our proof of concepts have shown us already about AI marking, here.
