2026 AI Conclave at IIT Madras
From our Founder’s Desk : Satyanarayanan Visvanathan
October has been a month of 𝐫𝐀𝐈𝐧 in Chennai this year. For Yashmia Consulting and Technologies, it has been a 𝐭𝐫𝐀𝐈𝐧 of AI activities starting with the launch of our AI services on 4th Oct ( https://lnkd.in/gnd-CCC2 ), followed by the hosting of a global webinar on titled RARE – Responsible AI for Responsible Enterprises on 10th Oct ( https://lnkd.in/gMaw3cbm ) Today on 22nd Oct Sashikala Viswanathan and I participated in the “Industry Conclave on AI Governance 2026” at Indian Institute of Technology, Madras. Hosted by Centre For Responsible AI (CeRAI) under the theme “AI Measurement and Governance”, the event was well attended with some 300+ participants
Balaraman Ravindran briefed about the journey of CERAI since its formation in Dec 2022. In his breezy talk, he explained how work has been going on in the space of responsible AI and building AI safe for a long time, much before the LLM based solutions muddled the space. He underscored the importance of right sized AI for India
Sarvapali (Gopal) Ramchurn started with the need of the hour viz. The antidote to AI Anxiety. He emphasised the importance of measurement in this and then Governance. His keynote address touched upon many important aspects such as – Jobs? Impact on future generations? Who to trust to fix?; How learning will be affected? How relationships are affected? ; For Governance to function, Measurements to work. The measurement challenges are complicated because humans are unique and society is a complex system
A series of papers were released by Dr Kamakoti, Director IIT Madras along with Prof Ravindran and the research related faculty and stakeholders
Dr Kamakoti started his address with the observation about the deep penetration of AI across all walks of life. He stated that any new technology will have good and the bad. Responsible Academic institutions will have to enhance the good and minimize the bad. He briefed about the five activities of IITM in this regard. He pointed out that we need to realise how AI can help at grassroots levels as well as how AI can help improve the quality of jobs. He made a powerful statement that India cannot afford to waste energy and if we make significant inferences with standard computing than heavy energy gulping systems – this frugal AI is vital
Mohan Kankanhalli deep-dived into the AI measurements by expanding on one theme viz. datasets and benchmarking. He narrated how their team secured the first place in the 2025 MLLM Jailbreak challenge (which he said was surprising 😊) and later a more modest sixth place. He briefed about the collaborative efforts that they did with far.ai towards red-teaming Anthropic Claude in terms of its safeguards. He gave insights about many of the work that they do such as GroMo dataset, SIMFONI dataset etc
