AI for AKM and Asesmen Nasional Prep

AKM (Asesmen Kompetensi Minimum) and Asesmen Nasional prep works best when it targets literacy and numeracy reasoning, not content recall, because the assessment measures how students read, interpret, and solve problems rather than how much they've memorized. AI can help by generating leveled reading passages, real-context numeracy problems, and AKM-style diagnostic questions that build those specific skills over time.

Quick Answer: AKM and Asesmen Nasional are not a pass/fail exam for individual students — they're a sample-based, PISA-style diagnostic of literacy and numeracy that schools use to target improvement. AI's best role is generating practice material in AKM's actual question formats, not generic drills, and helping teachers turn diagnostic gaps into focused weekly instruction.

That distinction matters more than it might seem. A teacher who treats Asesmen Nasional like the old Ujian Nasional — cramming content the week before — is preparing for the wrong thing entirely. This guide covers what AKM actually tests, where AI genuinely helps, and where a teacher's own judgment about a specific class still has to lead.

This kind of assessment reform is one piece of a much larger global picture — see AI in Education Around the World: A 2026 Regional Guide for how it's playing out across different countries.

What AKM and Asesmen Nasional Actually Measure

Asesmen Nasional (AN) is Indonesia's national education evaluation system, introduced by Kemendikbudristek (the Ministry of Education, Culture, Research, and Technology) in 2021 to replace Ujian Nasional, the earlier high-stakes exit exam. Unlike its predecessor, AN produces no pass/fail score for any individual student.

AKM: Literacy and Numeracy, Not Subject Recall

AKM tests two competencies — literacy and numeracy — using item formats closer to the OECD's PISA framework than a typical end-of-chapter quiz. Literacy items ask students to locate information, interpret meaning, and evaluate a text; numeracy items apply number sense, measurement, and data reasoning to real situations rather than isolated computation.

• Literacy spans informational and literary texts, at three cognitive levels: finding information, interpreting and integrating ideas, and evaluating and reflecting
• Numeracy covers number, algebra, geometry and measurement, and data and uncertainty, almost always framed as an applied word problem
• Neither section rewards a memorized formula or fact the way a traditional content exam does

Character and Learning Environment Surveys

Beyond AKM, Asesmen Nasional includes a Survei Karakter (Character Survey), which gauges dimensions of the Pancasila Student Profile, and a Survei Lingkungan Belajar (Learning Environment Survey), completed by teachers, principals, and students about school climate and teaching quality. Together, all three instruments build a fuller picture than a single test score ever could.

Sample-Based by Design, Not Individual Pass/Fail

AN is administered to a sample of students, typically in grade 5, grade 8, and grade 11, rather than every student in every grade. Its purpose is evaluating the school and the wider system, feeding results into Rapor Pendidikan (the ministry's education report card platform) — not producing a certificate any single student needs to pass.

How AN Differs From the Old Ujian Nasional

Asesmen Nasional didn't just rename Ujian Nasional — it changed what's measured, who takes it, and what the results are used for. Seeing the contrast side by side explains why old UN-style prep habits can actively work against AKM performance.

The practical consequence: a teacher who over-indexes on cramming specific content the week before an AN sitting is optimizing for a test that no longer exists. AKM rewards the ability to reason through unfamiliar material, a skill that only builds through repeated, varied practice over time. That's a genuinely different prep problem than a high-stakes, individually scored MCQ exam like AI for ECAT and Engineering Entry Tests presents, where every question directly affects one student's own result.

Why This Shift Changes How Prep Should Work

Moving from content-based UN prep to competency-based AKM prep is a bigger change than the acronym swap suggests, and it lands on top of a real, well-documented learning gap that makes the shift more urgent rather than less.

From Recall Drills to Reasoning Practice

A student who has memorized every formula in a textbook can still struggle with AKM, because the test asks them to apply reasoning to an unfamiliar scenario rather than reproduce a taught example. Effective prep now means practicing the skill of interpreting a new text or an unfamiliar word problem repeatedly, in varied contexts, not repeating drilled content.

The Learning-Loss Backdrop Makes This Urgent

The OECD's PISA 2022 results, released in December 2023, placed Indonesia's reading, mathematics, and science scores below the OECD average, continuing a gap that predates the pandemic and widened during it. The World Bank's learning poverty estimates (2022) found that more than half of Indonesian ten-year-olds could not read and understand an age-appropriate text — a foundational gap that AKM's literacy focus is specifically designed to surface early.

This is also why AKM prep should never be treated as exam cramming. INOVASI, an Australia-Indonesia government partnership focused on foundational literacy and numeracy, has spent years working directly inside classrooms on exactly this kind of skill-building — a sign that closing this gap takes sustained practice, not a pre-test sprint.

How AI Can Support AKM-Aligned Instruction

AI's clearest value here is generating practice material in AKM's actual formats — leveled passages, applied numeracy problems, and mixed-format diagnostic items — rather than generic multiple-choice drills that don't resemble what students will actually see.

Generating Leveled Literacy Passages

Say you teach a Grade 5 class with a wide reading-level spread. You could ask an AI tool for a short informational or literary passage at a specific reading level, paired with three question types matching AKM's cognitive levels: one that asks students to locate a detail, one that asks them to interpret intent, and one that asks them to evaluate or connect the text to their own experience.

Building Real-Context Numeracy Problems

AKM numeracy rewards applying math to a situation — a market transaction, a schedule, a simple dataset — rather than solving an equation in isolation. A teacher planning a Grade 8 numeracy lesson could request word problems built around everyday contexts, such as buying snacks, measuring a room, or reading a simple chart, at the exact competency level being taught that week.

Turning Rapor Pendidikan Gaps Into Weekly Focus

• Once a diagnostic report flags a weak area — interpreting texts, say, or data and uncertainty — a teacher can generate targeted practice sets on exactly that gap
• Mixing AKM's actual item formats (simple multiple choice, complex multiple choice with more than one correct answer, matching, short constructed response) into practice, instead of only single-answer multiple choice, familiarizes students with formats they'll actually encounter
• EduGenius can generate a set of leveled reading passages or numeracy word problems from a saved class profile, letting a teacher reuse the same grade-and-subject context across several practice sets instead of rebuilding it from scratch each time

None of this replaces a teacher's own read on which students need more scaffolding, or which passage topics will genuinely engage a specific class — AI drafts the raw material; the teacher still decides how to use it.

A Practical Weekly Routine

Building AKM-style skills works better as a steady weekly habit across the term than as a pre-assessment cram session, since literacy and numeracy reasoning develop gradually rather than overnight.

1. Pick one literacy focus and one numeracy focus for the week, based on diagnostic data or your own observation of where students struggle
2. Generate one short passage and one problem set at the right level, matched to AKM's cognitive levels rather than pure recall
3. Run it as low-stakes practice, not a graded test — the goal is repetition of the reasoning skill, not a score
4. Review answers together, focusing on why a wrong choice was tempting, not just which option was correct
5. Rotate topics weekly so students meet varied contexts — a story, a recipe, a bus schedule, a simple chart — rather than one repeated format
6. Revisit a skipped focus area at least once a month so nothing gets permanently dropped

Pro tip: Keep a simple running note of which AKM cognitive level — retrieve, interpret, or evaluate — each practice item targets. Over a term, this shows whether your practice has quietly drifted toward the easiest level, retrieval, which is the most common way well-intentioned AKM prep underdelivers.

This kind of steady, format-aware practice is a very different exercise from prepping students for one high-stakes sitting — compare it with a genuinely pass/fail context in AI Study Buddy for Matric (SSC) Board Exam Prep or AI for CUET Preparation in India, where exam-day performance is the entire point.

Supporting Different Classroom Contexts

More than 200,000 schools operate across Indonesia, spanning enormous variation in resourcing, class size, and how far Kurikulum Merdeka adoption has progressed — a single generic approach to AKM prep serves almost none of them equally well.

Large, Under-Resourced Rural Classrooms

A teacher managing a large class with limited materials has far less time per student for differentiated reading levels than a teacher in a well-resourced setting. The UNESCO Institute for Statistics has repeatedly flagged large class sizes and teacher shortages as structural features across many Southeast Asian public school systems, Indonesia included.

Generating a small set of leveled passages once, then reusing and lightly adapting them across several sessions, is a realistic way to bring differentiation into a resource-constrained room. The same large-class pressure shapes AI-assisted planning elsewhere too — see AI for Large Class Sizes in Kenya for how a different system handles a related constraint.

Private and Urban Schools Further Along Merdeka Adoption

Schools that adopted Kurikulum Merdeka earlier, often private and urban ones, are typically further along in shifting classroom habits toward competency-based instruction already. Here, AI's value shifts from introducing the format toward increasing the volume and variety of practice — generating passages across more genres and numeracy problems across more real-world contexts than one teacher could draft alone.

Multigrade and Composite Classrooms

Where one teacher covers multiple grade levels in a single room, a shared literacy theme with two or three difficulty tiers of the same passage lets every student practice the same skill at their own level simultaneously, without the teacher building three unrelated lessons from scratch. The same shared-theme, tiered-difficulty approach works for numeracy too — one applied word problem, restated at a simpler and a more advanced level, keeps the whole room working on the same underlying competency together.

Islands, Remote Districts, and Intermittent Connectivity

Indonesia's geography means a meaningful share of schools sit far from reliable internet access, particularly across the outer islands. A batch-generation habit — connecting once a week to produce several days of passages and problem sets, then working entirely offline afterward — fits this reality far better than a workflow that assumes constant connectivity.

Tools Teachers Can Use

You could use EduGenius to generate a batch of AKM-style literacy and numeracy sets for the week once your class's grade level and ability range are set in a class profile, then export the set as a PDF for printing where classroom internet access is unreliable. Its Bloom's Taxonomy alignment is a design feature meant to keep the cognitive level of a question visible, useful shorthand for checking whether a set tests recall or genuine reasoning.

This same everyday-connectivity constraint shows up well beyond Indonesia — see Offline and Low-Data AI Tools for Schools in South Africa for how a different system builds around it, or Best AI for Math Problems in 2026 (Benchmarked) for a closer look at how specific models handle numeracy problems.

Getting Started: A 30-Day Plan for AKM-Aligned Practice

Whether you're piloting this alone or coordinating with a grade-level team, the same first-month sequence works: confirm where students actually stand, build one repeatable practice habit, then expand once it runs smoothly.

1. Week 1: Pull your school's most recent Rapor Pendidikan report, or run a quick informal diagnostic if the school hasn't received one yet, to identify one literacy gap and one numeracy gap worth targeting.
2. Week 1: Draft a reusable prompt template that specifies grade level, AKM cognitive level, and topic, so every practice set you generate stays consistent in difficulty and format.
3. Week 2: Run your first weekly cycle — one passage, one problem set, reviewed together — with a single class before expanding to others.
4. Week 3: Adjust the template based on what felt too easy or too hard, and start rotating in the less-common AKM formats, like matching and short constructed response, alongside multiple choice.
5. Week 4: Share what's working with a colleague or grade-level team; a shared template saves everyone from rebuilding the same routine from scratch.
6. Ongoing: Revisit diagnostic data each term, since the specific gap worth targeting shifts as students improve.

Skipping straight to a full-department rollout is the most common mistake school leaders make with any new routine. Piloting narrowly with one class first catches format or difficulty problems before they multiply across a whole grade level.