AI IB classrooms are already here. At DSB International School in Mumbai, artificial intelligence is not a future consideration or a policy debate; it is active inside lessons, shaping how students receive feedback, develop critical thinking, and access the curriculum.
The principle guiding all of it is simple: technology must serve pedagogy, not the other way around. In an IB context, that means using AI to strengthen criterion-based assessment, deepen Approaches to Learning skills, and uphold the IB’s commitment to equity, integrity, and conceptual understanding. Not replace them.
Here is what that actually looks like in practice.
Personalised Learning That Still Holds Students to the Same Standard
One of the most useful things AI can do in an IB classroom is give students individualised feedback without lowering the bar everyone is being held to.
In DP English at DSB, students submit draft paragraphs and ask Gemini to identify where their analysis is descriptive rather than analytical in relation to Criterion A. The AI does not improve the writing. It highlights patterns and asks students to make decisions themselves. The work of analysing, revising and improving stays with the student. What changes is how quickly and specifically they can see where they are falling short.
This builds real ATL skills. Self-management, reflection, organisation, analysis and evaluation are all engaged in a way that transfers directly to summative assessment and to learning after school. Students are not working towards a model answer. They are learning to interpret criteria, monitor their own progress, and act on feedback deliberately.
What AI Does for Teachers
The honest version of this conversation includes what AI is doing for teachers, not just students.
At DSB, a small DP school where teachers often manage entire subjects alone, AI has reduced the administrative weight that eats into the time that actually matters. Identifying common misconceptions across a set of drafts, grouping students by feedback needs, and preparing differentiated scaffolds ahead of lessons, these are tasks AI handles well.
The result is that teachers spend more time in the conversations that cannot be automated. Conferencing. Targeted instruction. The human interactions that determine whether a student actually understands something.
Three subject examples show how this works in practice:
In DP Maths AI HL, Gemini identifies recurring technical errors in student explorations, such as inappropriate graphing windows or misread regression outputs. The teacher then groups students by conceptual need and uses class time for focused conferences where students justify their mathematical reasoning rather than correct technical mistakes the teacher has already found.
In DP Biology, Gemini scans IA drafts for common misconceptions in data processing and conclusion writing. With those patterns already identified, the teacher designs a targeted mini lesson on evaluating anomalies and uncertainty, and reserves one-to-one time for discussions about the scientific reasoning behind each student’s specific conclusions.
In DP History, students use Gemini to self-check whether their essay plans address all parts of the prescribed question before coming to class. The teacher then spends lesson time challenging students to justify their choice of case studies and historiography in relation to Criterion B, rather than correcting structural problems students could have caught themselves.
In all three cases, AI is handling the pattern recognition. The teacher is doing the teaching.
AI Literacy as an ATL Skill, Not a Tech Lesson
At DSB, AI literacy is not a standalone subject or a digital skills module. It is built into ATL development, which is exactly where it belongs.
Students are explicitly taught that AI outputs are not authoritative, neutral or complete. Learning tasks are designed so that evaluation, not production, is the goal. In Language and Culture, students use Gemini to generate multiple interpretations of a global issue, then work through structured prompts to interrogate what came back. Which perspectives are missing? What assumptions underpin this response? Where does the explanation oversimplify?
Students then compare AI-generated responses with course texts, teacher input and peer discussion. Assessment focuses on the quality of student judgment, not the polish of the final product. The learning happens in deciding what to keep, adapt or reject.
Inclusion, Multilingualism and Levelling the Playing Field
AI has made a real difference for students who might otherwise be locked out of parts of the DP curriculum.
In Language B and other subjects, students use Gemini to create bilingual glossaries, rephrase complex task instructions, or work through concepts at different levels of linguistic complexity. This lets students engage with demanding academic content while continuing to develop language proficiency. Expectations do not drop. Access improves.
For students with ADHD, tools like NotebookLM have made a particular difference. Students can upload teacher-approved sources and use the tool to generate structured summaries, concept maps, guiding questions or short videos. This reduces cognitive load without reducing academic challenge. In practice, students have sustained attention for longer, planned more effectively, and engaged more confidently in teacher conferences. The impact maps directly onto IB ATL self-management skills: organisation, time management and reflection.
Used intentionally, AI supports independence rather than advantage. It levels access rather than widening gaps.
Ethics, Integrity and the Traffic-Light System
None of this works without a clear ethical framework, and at DSB that framework is explicit and visible.
The school uses a traffic-light system to make expectations clear and developmentally appropriate. Green-zone AI use might include clarifying task instructions or generating planning questions. Before submitting work, students apply the traffic-light framework, review their AI use, and annotate their draft accordingly. A student might note that AI helped them organise ideas but that all analysis and writing decisions were their own.
Students understand that a red light means they cannot use AI to generate text they then pass off as their own. This reflection becomes part of the learning process itself, reinforcing academic honesty and transparency rather than treating it as an afterthought.
All AI-supported work receives teacher feedback. AI is treated as a supervised learning support, not a private space or shortcut. Students use school-managed Gemini accounts, with explicit teaching around data privacy, digital footprints and consent. Human oversight is present at every stage.
5 Principles for IB Schools Starting This Work
For other IB schools exploring how to bring AI into classrooms thoughtfully, DSB’s experience points to five things that matter:
Start with pedagogy, not tools. Clarify what learning problem AI is addressing before choosing a platform.
Anchor AI use in IB frameworks. DP criteria, ATL skills, and Approaches to Teaching provide the right foundation.
Teach judgment, not dependency. Design tasks where students must evaluate AI output, not accept it.
Make ethics visible. Clear guidance builds trust with students, staff and parents.
Start small and iterate. Pilot one use case, reflect on what happened, refine it, then scale.
The Bottom Line
AI in IB classrooms is not replacing teachers or thinking. At DSB, it is amplifying feedback, broadening access and deepening reflection. The challenge is not whether to engage with AI. It is how deliberately schools choose to do so.
When technology serves the IB’s values rather than competing with them, the results speak for themselves.
Source: IBO