Infosys founder N.R. Narayana Murthy didn’t mince words as he took aim at India’s tech scene, calling out the "fashion" of labeling ordinary software as artificial intelligence (AI).
Speaking at TiE Con Mumbai 2025, Murthy warned that what many companies parade as AI are nothing more than "silly, old programs" dressed up in hype.
Addressing a packed audience at TiE Con Mumbai 2025, Infosys founder N.R. Narayana Murthy expressed deep skepticism over the way AI is being misrepresented in India. “I think somehow it has become a fashion in India to talk of AI for everything. I have seen several normal ordinary programs touted as AI,” Murthy said, pulling no punches.
Murthy went on to explain the two fundamental principles of AI, cutting through the noise. “One, machine learning, which is a large-scale correlation. Therefore, it helps you based on a large amount of data, to predict,” he said.
“That is simple machine learning. The second is what is called deep learning. Deep learning imitates how the human brain works,” he added, clarifying the core mechanisms that differentiate true AI from mere code.
Elaborating on deep learning, Murthy said it can tackle “unsupervised algorithms”, unlike machine learning which relies heavily on supervised data. “While machine learning by and large handles supervised algorithms because you have to give a lot of data into that, deep learning uses the data to create new branches of programs or new conditions. And then it will be able to take decisions,” he explained.
“So unsupervised data, which uses deep learning and neural networks, has much greater potential to do things that mimic human behaviour better. But what I am seeing being called AI is silly, old programs,” Murthy said, breaking the illusion surrounding AI in India’s tech landscape.
Meanwhile, Infosys is working on its own AI models, focusing on Small Language Models (SLM) built with open-source components and proprietary datasets. The company has been pushing Generative AI (Gen AI) for specific industry applications and use cases, moving beyond superficial AI labels to build meaningful tools.