Best Place To BuildProf. Ravindran on The IIT Madras Playbook for Building AI Leaders | BP2B S1 Ep. 6
CHAPTERS
AI at IITM: From 6 students in RL to 35,000+ online learners
Prof. Ravindran contrasts the early days of teaching reinforcement learning at IIT Madras—when hardly anyone knew the term—with today’s massive demand across campus and online. He explains how IITM’s classes, NPTEL lectures, and the BS program scaled AI education far beyond the classroom.
Machine Learning vs Reinforcement Learning: Learning from sparse feedback
He gives an intuitive explanation of supervised machine learning as learning from labeled examples, then contrasts it with reinforcement learning where feedback is sparse and evaluative. Everyday skills like cycling, swimming, and walking illustrate why RL is a natural model for trial-and-error learning.
A pioneer’s path into AI: Chess, neural nets, and RL’s biological roots
Ravindran describes how early 1990s interest in chess-playing programs sparked curiosity about how “machines think.” After disappointment that neural networks had drifted from biological explanations, he found reinforcement learning through neuroscience experiments with monkeys and dopamine-like signals.
Multi-Armed Bandits: The slot-machine origin story and why it matters
He introduces multi-armed bandits as a classic decision problem with uncertain outcomes and multiple choices. The name comes from “one-armed bandit” slot machines, where you repeatedly choose actions to maximize reward under uncertainty.
Exploration vs Exploitation: The central dilemma in RL (and in life)
Using a foodie-versus-movies example, he explains why learning systems must explore options to discover value but also exploit known good choices to avoid wasted opportunity. The difficulty is deciding when to switch—too early risks suboptimality; too late wastes rewards.
“AI is the new CS”: Why AI is becoming its own discipline
Ravindran compares AI’s current maturity to computer science in the early-to-mid 1980s, when CS pulled away from electrical engineering. He argues AI draws deeply from psychology, control, economics, and math, so it can no longer be treated as just a CS subfield.
Origin of the Wadhwani School: Interdisciplinarity + scale + alumni support
He shares how IITM’s AI push began as an interdisciplinary effort recognizing algorithmic innovation happening across departments. Over time, degrees and centers grew large enough that creating a School became the natural organizational umbrella, accelerated by major alumni philanthropy.
Building AI centers like a startup: Teams, funding, KPIs—and inevitable attrition
He compares founding research centers/departments to startups: success depends on teams, fundraising, and staying focused on deliverables. Academia differs in output definition and incentives, and has built-in turnover because students graduate and leave.
Robert Bosch Centre (RBC-DSAI): A national template for interdisciplinary AI research
He traces the evolution from the Interdisciplinary Lab for Data Science into the Robert Bosch Centre, initially focused on network analytics and engineering-facing AI. The center scaled across departments, produced projects/papers/startups, and became a blueprint copied by other IITs.
From iBSC to healthcare AI: Biology, drug discovery, and space-station microbiomes
Ravindran highlights systems biology as another strong interdisciplinary thrust at IITM, applying AI to gene/protein networks and omics data. He notes work spanning drug discovery and microbiome studies, including research connected to the International Space Station.
Center for Responsible AI (CeRAI): Deploying AI safely in public services
He explains why “responsible use” matters as AI moves from demos to real-world deployments, especially in government and public-facing systems. CeRAI expands interdisciplinarity further by bringing in law, economics, sociology, and other perspectives beyond technologists.
AI’s interdisciplinary signal: Nobel recognition and AlphaFold’s field-level impact
The conversation turns to Nobel Prizes acknowledging AI-driven breakthroughs and how they reshape scientific practice. He discusses how AlphaFold transformed protein science by enabling structure databases, shifting what problems researchers can pursue.
IIT Madras “build culture”: CFI, Research Park, and leadership that removed boundaries
Ravindran describes how IITM’s innovation momentum comes from institutional DNA, leadership vision, and enabling infrastructure. He recounts CFI’s growth, the director-level push to break departmental silos, and the Research Park’s role in connecting work to industry and startups.
Before “going viral”: early web communities, forums, and online kinship
He shares personal stories of discovering the early web, creating a proto-blog food page, and building the Tamil Film Music page with large-scale discussion forums—an early social network before modern platforms. This segues into how online/offline relationships are blending for newer generations.
Future of work, human-AI coevolution, and why AGI is a “misnomer”
Ravindran argues AI won’t simply “take jobs,” but will change how every job is done—similar to how Excel became a baseline workplace skill. He also explains why “general intelligence” is poorly defined, and points to how AI can reshape human behavior (e.g., new Go strategies inspired by AlphaGo).
Designing the BTech in AI & Data Analytics (ADA): Ground-up curriculum for builders
He explains IITM’s new BTech in AI & Data Analytics as a curriculum designed from scratch rather than “CS + a few AI electives.” The program emphasizes math foundations, heavy programming, and building deployable systems—aimed at producing AI engineers ready for real-world software scale.
Get more out of YouTube videos.
High quality summaries for YouTube videos. Accurate transcripts to search & find moments. Powered by ChatGPT & Claude AI.
Add to Chrome