The AI class we are bunking
Self-driving cars, drones that deliver vaccines to remote villages, financial systems that adapt real-time. These are no longer in the realm of science fiction; they’re fast becoming a part of the global artificial intelligence landscape. Yet, many in India, including our educators, are unfamiliar with ‘agentic AI’—the specific kind of AI that underpins these technologies.
So, what is agentic AI? Think of it as a level up from traditional AI. These intelligent systems respond to data and can make independent decisions, learn from their environment, and interact with other intelligent agents. They don’t just process—they adapt, plan, negotiate and sometimes even disagree. In essence, they behave a bit like us.
The current AI curriculums in many of our institutions are rich in machine learning, computer vision and natural language processing. Impressive, yes. However, there’s a noticeable gap in preparing students to design systems that can reason, collaborate, and operate autonomously in unpredictable settings. This isn’t about just adding another topic to the AI syllabus. It’s about recognising a disruptive transition in how machines think and act—and ensuring India’s future engineers are ready for it.
Agentic AI is more than a collection of algorithms. It draws from diverse disciplines—cognitive science, decision theory, robotics and ethics, to name a few. It’s about creating intelligent entities that can make real-time decisions based on context, memory, internal goals and social cues. That is a complex task. And it requires a rethink of how we teach AI.
So far, most university courses focus on AI that works with static data sets—systems that are told what to do. What’s missing are courses on agents that figure things out for themselves. Our universities must train the students to master the principles behind multiple agents interacting—whether cooperating or competing—in shared environments. This is the backbone of applications like autonomous fleets, robot-assisted manufacturing and intelligent supply chains.
And let’s not forget ethics. Tough questions stare at us as machines begin to make decisions independently. How do we ensure fairness? What does transparency mean when machines independently change their strategies mid-task? These are not casual asides. They’re central to designing AI that earns human trust.
That said, change is in the air. Some IITs are developing real-world applications where students build autonomous systems to solve sustainability and urban infrastructure challenges. Indian industry players are moving beyond AI-powered dashboards to agentic platforms that reimagine how enterprises run. Startups, too, are rising to the occasion, building niche solutions that draw heavily on agentic AI models.
This surge of activity outside the classroom offers something our universities need to embrace—experiential learning. Startups, for instance, are perfect playgrounds for aspiring AI professionals. They offer the chaos and creativity that agentic AI demands. Unlike large corporations, startups pivot quickly, take risks, and break things. That’s precisely the environment where tomorrow’s AI innovators learn to thrive.
Yet, the foundation still lies in education. And here’s where Indian universities have a real opportunity. They must introduce courses that explore the inner workings of cognitive architectures—the models that allow agents to reason and make decisions like humans.
The curriculum must include lab experiments that let students build agents that learn by doing, not by being told. Educators must bring interdisciplinary perspectives from psychology, neuroscience, and ethics to explore the full implications of intelligent behaviour. Right now, such course offerings are missing from our university engineering education. What we need is mainstream exposure—early, accessible, and hands-on.
Does this all really matter for India? The answer, unequivocally, is yes. We do not want to produce coders anymore. We must shape the architects of the next digital age. And agentic AI is not some distant frontier—it’s already impacting defence, climate modelling, healthcare, and global supply chains. The country that trains its talent to lead in this space won’t just earn patents and profits; it will mould global norms, influence technological standards, and perhaps even set the ethical tone for autonomous systems worldwide.
For Indian higher education institutes, the message is clear. Teaching our students to master basic AI isn’t enough. If our higher education institutions are serious about preparing students for the age of intelligent machines, they must go far beyond teaching the mechanics of algorithms or neural networks. We must enable students to think critically about how autonomous systems learn, adapt, and interact, often in unpredictable environments.
The curriculum must integrate this technical know-how with a clear grasp of social impact and ethical design. This approach requires an interdisciplinary outlook. Designing an AI model is one thing; anticipating how it behaves when faced with conflicting values in the real world is quite another. What should an AI system do when choosing between efficiency and fairness? Who is accountable when algorithms go wrong? Questions like these are the edge cases of tomorrow’s challenges, and our students must be ready to address them.
It’s a rare confluence of need, talent, and timing—Indian educational institutes must act—to become a launchpad for creative disruption in intelligent autonomy. That journey can begin right now in Indian classrooms, so long as we prioritise relevance over being tied to legacy systems.
Mamidala Jagadesh Kumar
Former Chairman of University Grants Commission and Vice Chancellor of JNU
(Views are personal)