Last week, we carried a story on the

importance of math in AI

. We received good feedback on that, so we decided to carry forward this conversation, and once again underline how crucial the subject is becoming. For India to become really good in AI, we have to up our math and statistics capabilities. That, experts say, is the only way for us to come out with DeepSeek-like innovations.
Shyam S Kamath, professor of math in NIT (National Institute of Technology) Karnataka, notes that algorithms that are at the root of all AI are nothing but mathematical equations. It’s all logic, he says, adding that the only way to handle vast amounts of data is through math.
Data scientists are at the heart of this. They create knowledge from data through quantitative and programming methods – enabling the image processing, text categorisation, and predictions that we see AI models doing. Rudramuni B, former head of Dell’s R&D centre in Bengaluru, and now an associate partner in a VC firm, says the data scientist has to be a better statistician than a software programmer, and also a better engineer than a statistician. “The data scientist applies mathematics in solving problems through software,” he says.
Even if you have a pre-built AI model, if something in the environment changes because of which the accuracy of the model changes, you may have to change the underlying algorithm. And that is not like changing a feature on a phone by programming, says Rudramuni.

“It’s a bunch of equations. You can’t change it without understanding the math. If accuracy has to be 99%, it cannot be done at the code level, you have to dig into the algorithm, understand the basis of the algorithm. A computer science student may not understand it,” he says.
It is with this understanding that NITK Surathkal, for instance, offers its BTech in computational & data science programme through the math department and not the CS/IT departments. Kamath recollects that when the idea of doing it under the math dept was first mooted a few years ago, he took feedback from professors in IISc and others, and there was good support. He says the education ministry also wholeheartedly supported the course, and today there’s big demand for it.
Rudramuni says most engineering colleges currently produce programmers from CS/IT departments in the name of AI, ML, data science engineering, thereby diluting the importance of math. This, he says, is detrimental to India’s ambitions in AI.