Picture an engineering student in a tier-III town, stuck on a Python script late at night with no teacher to call. Until recently, her options were limited to generic fixes: rewatching a YouTube video or digging through online forums that do not adapt to the learner’s needs. Artificial intelligence (AI) is changing that, as new tools begin to reshape how students learn and access support.
Companies including PhysicsWallah, Coursera and Scaler are embedding AI into their offerings, exposing gaps in one-size-fits-all learning and forcing the industry to rethink its model.
Tools such as NotebookLM can turn a textbook into a personalized podcast or slideshow, while others build study modules that adjust difficulty based on how a student performs in real time. These are services edtech platforms have long offered as part of paid programmes, but AI is now delivering them faster and often at little to no cost, putting pressure on existing models.
Learning your way, your pace
The deepest problem AI is solving is indeed the straitjacket approach to learning. “The real problem with traditional education—online or offline—isn't content quality. It's that every learner gets the same path, regardless of where they are,” said Amar Srivastava, chief executive and chief product officer at Scaler. "A first-generation coder from a tier-III town and an IIT graduate revisiting fundamentals have completely different gaps, paces, and anxieties. A YouTube playlist can't see that difference."
What AI changes, Srivastava argues, is the feedback loop. Instead of the slow cycle of learning, taking a test, and getting a score days later, the system works in minutes: a student attempts a problem, gets a specific hint, retries, and moves to a harder variation. "AI also generates queries for them to critique—because in real workplaces, you'll spend as much time reviewing AI-generated code as writing your own. We're training that judgment muscle from day one," Srivastava added.
NxtWave takes the access argument further. The platform, which delivers tech education in regional languages, reaches learners across 650-plus districts and 11,300-plus pincodes across India. “Students don't fail because they aren't smart enough. They fail because the learning system wasn't updated as the industry demands,” said Rahul Attuluri, co-founder and chief executive of NxtWave. "A student in Visakhapatnam trying to understand data structures in English, at the same speed as someone in Bengaluru, who's been coding since school, that's not a level playing field."
Established platforms are responding by building AI into their own products. Physics Wallah has rolled out a range of tools, including Ask AI, a round-the-clock doubt resolution service that has answered nearly 3 million queries since September 2025, and an AI grading system that has evaluated over 2 million written answers.
"Our approach has been to build with AI, not compete against it," Physics Wallah said in response to Mint's emailed queries, adding that teachers remain central to outcomes in competitive exam preparation where discipline and motivation matter as much as content.
Edtech's AI edge
Coursera, which works with universities and employers to offer accredited courses, points to measurable results. Its AI tutor draws on actual course content, rather than the open internet to answer questions, and 98% of learners in India say it's helped.
Larsen & Toubro, which uses Coursera for staff training, reported a 25% improvement in understanding of complex topics. More than 3,500 of the 6,100 courses built using Coursera's AI authoring tool have come from Indian institutions, including Symbiosis International University, which used it to build interview preparation courses tied to placements. "AI is not replacing structured learning; it’s reshaping how it is delivered and experienced," Coursera said in response to Mint's emailed queries.
This is also where edtech platforms believe they hold an edge over standalone AI tools. Models that draw from the open internet are prone to hallucinations—producing answers that sound authoritative but are factually wrong. Platforms running on vetted, proprietary course content do not carry the same risk. Coursera Coach, for instance, answers questions grounded in the actual course a learner is enrolled in, not a general web crawl.
Even Google is moving in this direction. The company this week announced that its Gemini AI now offers free full-length practice tests for JEE Main and NEET UG, built on vetted content from Physics Wallah and Careers360. "This helps ensure that you are not just practising—you are preparing with material that more closely resembles what you will see on test day," Google said in a blog post. Its NotebookLM, meanwhile, is positioned as a tool that works on top of educator-created content rather than around it. "NotebookLM relies on and amplifies the expert content created by educators to support their vital role in student learning," a Google spokesperson said in response to emailed queries.
Will human teachers still matter?
The question of what this means for teachers—particularly the thousands who have built livelihoods on YouTube—sits uncomfortably beneath all of this.
The platforms state AI does not replace educators, and the argument has merit. A mentor who has interviewed hundreds of candidates for technical roles carries knowledge no model can replicate. An instructor who can tell that a student's confusion runs deeper than the current topic requires a human read. "AI handles the what and the how of learning. Teachers and mentors handle the why keep going," said Attuluri of NxtWave.
But the platforms are also candid about who is under pressure. "Educators whose entire model was built on recorded content, without community, without personalization, without a human layer beyond the video itself. For them, the headwinds are real and growing," Attuluri added.