When a student is asked to write an essay today, an AI tool can produce a reasonably polished answer within seconds. But a more fundamental question follows: does the essay reflect the student’s thinking, or simply the machine’s ability to generate convincing prose?
A Mumbai Class 12 student is attempting to tackle that question not by banning AI in schools, but by designing a framework for using it without allowing students to outsource the learning process.
For the past two years, Vedant Choudhury has independently developed VC Beyond, a non-profit platform comprising an AI policy framework, a guided learning tool and a free curriculum aimed at helping schools introduce AI in a structured way.
“I wanted to prove that schools can have AI tools without compromising students’ education,” Choudhury said. “Moreover, I wanted to prove that a student is capable of building both the tools and the rules.”
Moving beyond the AI-or-no-AI debate
The growing availability of generative AI has made banning the technology in classrooms an increasingly limited response. Students are likely to encounter AI throughout higher education and their careers, making the ability to use it responsibly as important as knowing when not to use it.
The challenge, therefore, is shifting from detecting whether AI was used to understanding how it was used and what the student actually learned.
The OECD’s 2026 review of AI in education highlights a related concern: students can produce better writing with AI assistance without necessarily acquiring the underlying skills. UNESCO, meanwhile, has advocated critical AI literacy supported by age-appropriate restrictions.
VC Beyond approaches the problem through three components: AIFA, SochMap and AIDE.
Turning AI policy into something schools can implement
AIFA is the platform’s AI policy framework. Instead of asking individual teachers to determine what constitutes acceptable AI use, administrators select parameters such as the education board, grade, role and language to generate customised guidelines.
The framework defines when AI can be used, when its use should be disclosed and which tasks students should complete independently.
The approach also changes according to age. Younger students use AI under greater supervision, while older students receive more autonomy along with greater responsibility for disclosure. Separate handbooks provide guidance for students, teachers and administrators.
The objective is to make AI governance part of the school’s operating framework rather than leaving it to individual teachers to interpret.
What if AI helped students think rather than write?
The second component, SochMap, takes a different approach to assignments.
Instead of asking AI to generate a finished essay, students build a map of the topic, connect ideas, examine sources and progressively develop their arguments. The tool provides tips and feedback during the process, but does not simply hand over a completed essay.
The underlying principle is straightforward: the difficult part of using generative AI is no longer producing text. The bigger educational risk is skipping the thinking that should have produced that text.
SochMap is designed to keep students involved in research, reasoning and decision-making while using AI as an assistant rather than an author.
Building AI literacy into the curriculum
The third component, AIDE, is a free 20–30-hour curriculum designed to build practical AI literacy.
Younger students learn how the technology works through simple tasks, while older students undertake projects involving AI tools and verification of their outputs.
Each course culminates in a capstone project that is live and deployed. The curriculum uses Indian examples, is designed to work on smartphones and is available in both English and Hindi.
That practical approach is intended to move AI education beyond simply teaching students how to write better prompts.
From workshops to school pilots
Choudhury’s work has also been shaped by his experience conducting AI workshops. He has delivered more than 20 workshops to over 500 participants, including CEOs and business leaders, and served as a teaching assistant during an AI training programme conducted in India by Singularity University for more than 200 CEOs.
Those experiences, he says, demonstrated how rapidly AI adoption was spreading—and how little many users understood about the technology when they first encountered it.
VC Beyond is still an early-stage initiative. Its framework has been piloted in two schools and refined based on feedback from teachers and other stakeholders. B.D. Somani International School has committed to using the framework.
The next stage is a broader pilot involving the policy framework, teacher training, workshops, curriculum and guided learning tools integrated into assignments during a regular school week.
The bigger question for schools
The challenge facing schools may ultimately be less about whether a student used AI and more about whether the student can explain what they did with it.
A student can submit an AI-generated essay. A much more demanding assessment is to ask the student to explain the choices behind the argument, the evidence used, the sources consulted and how the final answer evolved.
That distinction could become increasingly important as generative AI makes polished output almost effortless.
VC Beyond’s central proposition is therefore not that schools should find better ways to police AI. It is that schools may need to redesign learning and assessment around a world where AI is already available.
The question is no longer simply whether students should use AI. It is whether education can teach them to use it without giving up the thinking that education is supposed to develop.