Efficacy, in the Oxford Rubric, asks whether AI use demonstrably enhances teaching quality and learning outcomes — not just operational efficiency. It requires schools to define an educational purpose before adoption, gather evidence of impact, and actively guard against the risk of over-scaffolding students out of the productive struggle that genuine learning requires.

Time-saving is the headline promise of almost every AI tool sold to schools. Marking faster. Lesson planning faster. Admin faster. And time saved is genuinely valuable to teachers who are already stretched.

But the Oxford Rubric's second criterion — Efficacy — asks a harder question: is the AI use actually making teaching better and learning deeper? Not just quicker. And the answer, for a surprising number of tools currently in use in secondary schools, is: we don't really know.

What Does the Oxford Rubric Mean by Efficacy?

The rubric defines Efficacy as whether AI use demonstrably enhances teaching quality and learning outcomes, rather than merely increasing efficiency or introducing novelty with little benefit to pupils.

The word "demonstrable" is doing a lot of work in that definition. It's not enough for an AI tool to feel useful, or for teachers to report that it saves them time, or for vendor case studies to claim impressive results. The Efficacy criterion requires evidence gathered within the school's own context, evaluated against educational metrics — not just technical performance indicators.

The rubric is also direct about the risk of over-scaffolding: "AI should serve learning, not displace the pedagogical relationships, struggle, reflection and practice through which learning occurs." For secondary teachers, this is a familiar tension. Homework tools, AI writing assistants, and adaptive learning platforms all raise the same question — at what point does support become substitution?

The Over-Scaffolding Problem in Secondary Schools

Over-scaffolding happens when an AI tool makes a task so easy that the cognitive effort required to complete it — the effort that produces learning — is removed. A student who uses an AI tool to generate the first draft of every piece of writing may produce more polished work in the short term while developing less capacity to write independently over time.

This is not a hypothetical risk. It's a well-documented pattern in educational technology more broadly, and generative AI accelerates it significantly. The Oxford Rubric's Efficacy criterion makes it a formal evaluation requirement: before deploying any AI tool with students, schools need to ask whether it risks narrowing learning, reducing cognitive effort, or undermining assessment integrity.

For classroom teachers, this translates into a design question as much as a procurement one. How is the AI being used in the lesson? At what stage does it enter the process? Is it supporting the development of skills, or doing the work that would have developed them?

What Appropriate Efficacy Looks Like

The Oxford Rubric's indicators of appropriate Efficacy practice are worth knowing in full:

• The educational purpose is defined before adoption — not after the tool has been purchased.
• AI use is explicitly aligned with sound pedagogical principles.
• Evidence is gathered on impact on learning and teaching quality.
• Impact is evaluated using educational metrics, not just technical ones.
• AI supports feedback, insight, or differentiation without replacing the learning process itself.
• Use is regularly reviewed against educational outcomes, not novelty.

Compare that with the red flags: AI adopted primarily for efficiency or convenience; no evaluation of learning impact; cognitive demand reduced rather than enhanced; and assessment integrity undermined.

The Questions Secondary Schools Should Be Asking

• What is the specific educational purpose of this AI use? Can we state it clearly before we start?
• Is there peer-reviewed or independently verified evidence that this type of AI tool improves learning outcomes?
• How will we know if it's working? What are our educational success metrics?
• Does the way this tool is being used preserve the productive struggle and cognitive effort that learning requires?
• Are we evaluating this regularly, or assuming the initial case for it will hold indefinitely?

That last point matters more than it might seem. Educational evidence shifts. A tool that seemed promising in early trials may show different results when scaled, or when used with different student populations. The Efficacy criterion requires ongoing review — not a one-time procurement decision.

Efficacy and the Other Oxford Rubric Criteria

Efficacy doesn't sit in isolation. An AI tool can be safe (in terms of data protection and safeguarding) while still failing the Efficacy criterion if it doesn't demonstrably improve learning. And a tool that improves some measurable outcomes may still fail the Agency criterion if it does so by reducing student independence.

The Oxford Rubric requires all five criteria to be Green before a use case is considered appropriate. Efficacy is often the criterion that takes the most ongoing work — because evidence needs to be gathered, reviewed, and acted upon over time. But it's also the criterion most directly connected to the core purpose of a school: the quality of teaching and the depth of learning.

Frequently Asked Questions

What does Efficacy mean in the Oxford Rubric?

Efficacy is the second criterion of the Oxford Rubric. It asks whether AI use demonstrably enhances teaching quality and learning outcomes, rather than just increasing operational efficiency. It requires schools to define an educational purpose before adoption, gather evidence of impact, and guard against over-scaffolding — where AI does so much for students that it reduces the productive struggle through which real learning happens.

How do schools measure the efficacy of AI tools?

Schools should measure AI efficacy using educational metrics — such as improvements in student attainment, quality of feedback, depth of teacher planning, or student engagement — rather than technical performance indicators like response speed or usage volume. The Oxford Rubric requires this evidence to be gathered within the school's own context and reviewed regularly.

Can an AI tool save time and still fail the Efficacy criterion?

Yes. The Oxford Rubric explicitly distinguishes efficiency gains from educational gains. A tool that saves teachers time while reducing the quality of feedback students receive, or that makes it easier for students to complete tasks without developing underlying skills, may be efficient without being efficacious. Both matter, but Efficacy is the educational standard the rubric is designed to protect.