AI Can Give Answers, But It Cannot Replace Understanding: IIT Kanpur National Teacher Award Winner Sankar Prasad Rath on How Students Should Use AI
AI in Education: As generative artificial intelligence becomes a routine part of student life, IIT Kanpur Professor Sankar Prasad Rath has emphasised a principle that increasingly matters across schools and universities: technology can support learning, but it cannot substitute for deep conceptual understanding.
Professor Rath, from IIT Kanpur's Department of Chemistry, was recently conferred the National Award to Teachers 2026 for his contributions to teaching, research and institutional service.
His education philosophy places strong emphasis on fundamentals, curiosity, research and the relationship between classroom knowledge and real scientific problems.
Who Is Professor Sankar Prasad Rath?
| Particular | Details |
|---|---|
| Institution | IIT Kanpur |
| Department | Chemistry |
| Joined IIT Kanpur | 2004 |
| Full Professor since | 2014 |
| Major field | Bio-inorganic chemistry |
| Research papers | More than 165 |
| PhD students supervised | 27 |
| Recognition | National Award to Teachers 2026 |
Why Conceptual Understanding Matters More in the AI Era
Generative AI can now:
- Explain equations
- Summarise chapters
- Generate code
- Solve numerical problems
- Create study notes
- Suggest research ideas
But the system producing an answer does not transfer understanding automatically to the student reading it.
A learner may copy a correct solution without understanding why the method works.
That distinction becomes critical in subjects such as mathematics, physics, chemistry and engineering.
AI Answer vs Student Understanding
| AI Can Often Do | Student Still Needs to Do |
|---|---|
| Produce an explanation | Understand the principle |
| Solve an example | Apply concept to a new problem |
| Generate code | Understand logic and debug it |
| Summarise research | Evaluate evidence |
| Suggest an answer | Verify whether it is correct |
Why AI Can Be Wrong
Large language models generate responses by identifying patterns in data and context.
They can produce highly convincing explanations that contain:
- Incorrect facts
- Invented references
- Mathematical errors
- Outdated information
- Misinterpreted questions
A student without conceptual knowledge may not recognise those errors.
The stronger the learner's fundamentals, the more effectively the learner can use AI.
How Should Students Use AI?
A better learning workflow is:
- Attempt the problem independently.
- Identify exactly where you are stuck.
- Ask AI for an explanation or hint.
- Compare it with textbooks or reliable sources.
- Solve the problem again without AI.
- Explain the answer in your own words.
Use AI as a Tutor, Not an Answer Machine
Students can turn AI into a learning tool by asking questions such as:
- “Explain this concept at three difficulty levels.”
- “Give me a hint but don't solve the problem.”
- “Ask me five questions to test whether I understand this chapter.”
- “Find the mistake in my reasoning.”
- “Create a new problem using the same concept.”
This encourages active learning rather than copying.
Professor Rath's Teaching Record
IIT Kanpur says Rath has developed the undergraduate elective Inorganic Molecules, Materials & Medicine, connecting fundamental chemistry with biological and everyday processes.
He has also taught preparatory courses intended to strengthen the academic foundation of students from disadvantaged backgrounds.
That focus on foundations directly relates to the wider AI-learning debate.
Research and Teaching Do Not Have to Compete
After receiving the National Award, Rath said excellence in research and commitment to teaching strengthen one another.
His own research examines areas such as:
- Multiheme proteins
- Bio-inspired catalysis
- Chemical biology
- Medicinal inorganic chemistry
- Supramolecular chirogenesis
Research experience can make classroom teaching more powerful because students see how theoretical concepts become scientific investigation.
Should Schools Ban AI?
A complete ban can be difficult to enforce and may prevent students from learning how to use an increasingly important technology responsibly.
A more practical approach is to define:
- When AI is allowed
- When independent work is required
- How AI assistance should be disclosed
- What forms of copying constitute misconduct
- How students should verify AI output
AI and Exams
Traditional examinations test whether the student can independently retrieve and apply knowledge.
If an AI system completes all homework outside the exam hall, the student may appear productive without developing the ability needed during assessment.
That creates what could be called an AI dependency gap.
Why Fundamentals Become More Valuable, Not Less
When calculators became common, arithmetic did not disappear from education.
When search engines arrived, understanding did not become unnecessary.
Likewise, generative AI changes which tasks students perform manually, but it increases the importance of:
- Critical thinking
- Question formulation
- Verification
- Conceptual understanding
- Judgement
What Teachers Can Do Differently
Teachers can design assessment around reasoning instead of easy-to-copy outputs.
Examples include:
- Oral explanation
- In-class problem solving
- Project defence
- Comparison of alternative solutions
- Reflection on AI-generated errors
AI in Education FAQs
Can AI replace teachers?
AI can support tutoring and explanation, but education also requires mentorship, assessment, judgement and human interaction.
Should students use ChatGPT for studying?
AI can be useful when used for explanation, practice and feedback rather than simply copying answers.
Why is conceptual understanding important?
It allows students to apply knowledge to unfamiliar problems and recognise incorrect AI output.
Who is Sankar Prasad Rath?
He is a Chemistry professor at IIT Kanpur and a recipient of the National Award to Teachers 2026.
Final Takeaway
AI may become one of the most powerful educational tools ever created, but its value depends on how students use it.
A learner who understands the concept can use AI to move faster. A learner who does not understand the concept may simply copy mistakes faster.
The future of education is therefore unlikely to be AI versus teachers or AI versus textbooks. It will be students with strong fundamentals learning how to use AI without surrendering the thinking process.










