Chalkboard in a Mumbai public school classroom with a glowing AI chatbot cursor stenciled next to a half-solved math problem

Mumbai Study: 80% of Grade 9 Students Know AI, Few Understand It

Eight in ten ninth-graders in Mumbai’s public schools say they know what artificial intelligence is. Far fewer can explain how it works, and a growing share already lean on it to finish their homework. That is the central finding of a joint study released this week by the Salaam Bombay Foundation and NMIMS Mumbai, based on survey responses from 1,050 Grade IX students across 20 public schools and focus groups with 12 teachers. The report, dated April 2026, lands just three weeks after the Centre rolled out its first mandatory AI curriculum, and it argues that older students sit in a policy blind spot the rollout does not yet close.

The study’s sharpest claim is not that teenagers use AI. It is that they use it alone. Students in the sample reach for chatbots to solve math problems, translate passages between English, Hindi and Marathi, and draft assignments, most of them without a teacher looking over their shoulder and without a curriculum telling them when a tool is helping and when it is quietly doing the thinking for them.

What 1,050 Mumbai Teenagers Told the Researchers

The headline number, 80 percent AI awareness, came almost entirely from exposure rather than classroom instruction. Students named ChatGPT, Gemini, Meta AI inside WhatsApp and the voice assistants on their parents’ phones. Few could describe what a large language model is, how it is trained, or why it sometimes invents facts.

The report ties three specific behaviours to that shallow base of knowledge. Pupils use AI to crack algebra and geometry problems, to translate reading material, and to complete written homework. The authors argue each of those uses, left unguided, chips away at critical thinking, creativity and the back-and-forth between teacher and student that Grade IX is supposed to build.

The sample is small by national standards but deliberately focused. All 20 schools are public, drawn from different wards of Mumbai, and the 12 teachers who joined focus groups cover a spread of subjects and neighbourhoods. In other words, the findings describe the children the city’s private-school ecosystem tends to leave out of its edtech pitches.

The Quiet Gender Gap Inside the Data

Boys reported higher AI knowledge than girls, and they also reported leaning on AI more heavily to finish schoolwork. That pattern mirrors the broader digital divide captured by the Annual Status of Education Report 2024, which found 82 percent of 14 to 16-year-olds nationally can use a smartphone, but boys still enjoy greater independent access than girls.

Unequal access compounds unequal confidence. If girls in Grade IX are using AI less often and with less fluency now, the gap in who can prompt, audit and critique these tools later will be harder to close than a gap in awareness.

Why Grade IX Falls Through the Policy Crack

The timing of the Mumbai report is what gives it bite. On April 1, 2026, Union Education Minister Dharmendra Pradhan launched the Computational Thinking and Artificial Intelligence curriculum formally notified by the Department of School Education and Literacy in October 2025. The rollout is sweeping in one sense and narrow in another. It covers Classes 3 to 8 across all CBSE-affiliated schools from the 2026-27 session, embedding AI inside maths, science, social science and language teaching rather than as a stand-alone subject.

Grade IX is not in that list. Neither is Grade X. The students in the Salaam Bombay Foundation survey are already past the gate CBSE just built.

That gap matters because of how the curriculum is structured. CBSE’s own April 2026 training circular anchors teacher preparation in seven themes, from AI readiness and interdisciplinary integration to ethics and the role of mathematics in computational thinking. None of that scaffolding reaches a Mumbai ninth-grader this year, even as the same ninth-grader is already asking Gemini to translate her civics chapter.

What the GATEE Framework Actually Proposes

To plug the gap, the report pitches a framework with an unwieldy name and a clear premise. GATEE, short for Gateway to AI for Transformative and Equitable Education, argues that AI literacy should be treated as infrastructure, not enrichment.

Its recommendations fall into four buckets:

  • A standardised AI curriculum that extends past Class 8 and reaches the secondary years the CBSE rollout currently skips.
  • Foundational AI literacy built around critical thinking, not just tool use, so students can question the outputs they already trust.
  • Equitable access and digital infrastructure in public schools, where bandwidth, devices and electricity remain uneven.
  • Mandatory, practical training for teachers, backed by clear policy guidance on when AI belongs in a lesson and when it does not.

The framing echoes the UNESCO AI Competency Framework for Students, released in 2024 and updated in January 2026, which organises student learning around a human-centred mindset, ethics, technical foundations and AI system design. It also sits alongside the AI Samarth framework published by Central Square Foundation with IIT Madras in May 2025, which is already influencing NCERT’s curriculum thinking.

Teachers Without a Playbook

The 12 teachers who joined the study were not hostile to AI. They wanted a balanced mix of traditional teaching and AI tools. What they said they did not have was training, guidelines, devices, staffing or time.

That is the quieter story in the data. A Grade IX classroom in Mumbai often runs with a single teacher, a packed syllabus, and a handful of shared devices. Ask that teacher to detect ChatGPT-written homework, teach prompt literacy and flag algorithmic bias at the same time, and the maths stops adding up.

“The findings show that student engagement with AI is largely unguided, posing potential risks to learning if left unaddressed. This places an important responsibility on higher education institutions to lead the way by developing AI literacy modules, supporting teacher capacity building, and ensuring that AI becomes a tool for strengthening critical thinking rather than replacing it,” said Dr Meena Galliara, Director of the Jasani Center for Social Entrepreneurship and Sustainability Management at SBM, NMIMS.

Gaurav Arora, Chief Growth Officer at the Salaam Bombay Foundation, framed the findings as a starting line rather than a verdict. “The first step was to understand the prevalent usage basis, which we will now align with our programmes keeping in mind the National Education Policy as well,” he told reporters, adding that the foundation wants to scale the work through government partnerships so that more public-school students gain “equitable access to AI literacy.”

How the Numbers Stack Up Against the Rest of India

Mumbai is not an outlier. The Comprehensive Annual Modular Survey 2022-23 from the Ministry of Statistics and Programme Implementation reported that more than 95 percent of Indian youth can operate a mobile phone. Central Square Foundation’s Bharat Survey for EdTech 2023 found a smartphone in 85 percent of households.

Those two figures explain why the Salaam Bombay Foundation data reads less like a shock and more like a forecast of what the rest of the country’s public-school system will report if anyone bothers to measure it. Exposure runs ahead of instruction by a wide margin, and the tools only get more capable.

The table below lines up the major frameworks now competing for space in Indian classrooms.

FrameworkSourceTarget GroupStatus as of April 2026
CT and AI CurriculumCBSE, Ministry of EducationClasses 3 to 8Mandatory rollout from 2026-27
AI SamarthCentral Square Foundation, IIT MadrasMiddle schoolReference framework, May 2025
AI Competency FrameworkUNESCOStudents and teachers, globalPublished 2024, updated January 2026
GATEESalaam Bombay Foundation, NMIMSPublic secondary schoolsProposed, April 2026

What Comes Next for Mumbai’s Public Schools

Arora told reporters the Salaam Bombay Foundation plans to align its AI literacy programming with the National Education Policy and seek direct partnerships with the Brihanmumbai Municipal Corporation and the Maharashtra state government. NMIMS, in turn, has pitched itself as the higher-education anchor that trains teachers and builds modules for schools too small to do either.

Whether that partnership clicks will decide a lot. If it does, the ninth-graders in this study could be the last Mumbai cohort to meet AI without a teacher in the room. If it stalls, the gap between an eighth-grader trained under the new CBSE framework and a tenth-grader left to figure out Gemini alone will harden into something the next survey has to explain.

Frequently Asked Questions

What did the Salaam Bombay Foundation and NMIMS AI report actually measure?

The study surveyed 1,050 Grade IX students across 20 public schools in Mumbai and ran focus groups with 12 teachers from different parts of the city. It measured student awareness of AI, how often they used AI tools for schoolwork, and how prepared teachers felt to guide that use.

How are Mumbai students using AI in school?

They most often use AI to solve math problems, translate between languages and complete homework assignments. The report warns that this unguided use reduces critical thinking, creativity and direct interaction with teachers.

What is the GATEE framework?

GATEE stands for Gateway to AI for Transformative and Equitable Education. It recommends a standardised AI curriculum, stronger foundational AI literacy, equitable digital infrastructure in public schools, and mandatory practical training for teachers.

Does the new CBSE AI curriculum cover high school students?

Not yet. The curriculum launched on April 1, 2026, covers Classes 3 to 8 from the 2026-27 academic session. Secondary classes, including Grades 9 and 10, are outside the current rollout, which is one of the gaps the Mumbai report highlights.

Why does the report flag a gender gap?

Boys in the sample reported higher AI knowledge and greater dependence on AI tools than girls. The authors argue that unequal early fluency can widen later gaps in who can evaluate, audit and design AI systems, not just use them.