Vinod Bhat
New Delhi: The Research Symposium on AI & Its Impact, held as part of the India AI Impact Summit 2026 on 18th February 2026 the third day of the summit, convened leading researchers, policymakers, technologists and industry leaders to explore the transformative role of Artificial Intelligence across Science, Governance, Industry and Society. Designed as a platform to bridge the gap between frontier research with real-world application, the Symposium examined how AI can drive scientific breakthroughs while remaining aligned with public interest, safety and inclusive growth.
In his Special Address, Shri Ashwini Vaishnaw, framing India’s AI journey around practical deployment and population-scale impact, stated “Interacting with thousands of young people at the ongoing AI Expo, I was struck by their optimism about the future. That confidence has made me hopeful about a new chapter for our country and for the world. In India, our focus is on AI at the edge, AI that solves real-world problems, improves enterprise productivity, and addresses population-scale challenges in healthcare, agriculture, and climate change. This symposium is an opportunity to shape that future responsibly, and I urge leaders here to offer concrete ideas on how to make AI safe and truly beneficial for humanity.”
Highlighting AI’s potential to advance science and medicine, Sir Demis Hassabis, Co-Founder & CEO, Google DeepMind, said “We are at a threshold moment where artificial general intelligence (AGI) is on the horizon. AI will be one of the most transformative technologies in human history, with extraordinary potential to advance science, medicine, and human health but it also carries real risks. Because this technology will affect the whole world, international dialogue and cooperation are essential to ensure its benefits are shared and its dangers responsibly managed.”
Setting the academic context, Prof. P.J. Narayanan, Former Director, IIIT Hyderabad, underscored the rapid rise of AI within global scientific discourse and the need to engage diverse perspectives on both opportunity and risk. He highlighted the Symposium’s carefully curated structure, including plenary keynotes, research dialogues on frontier AI questions, Global South-focused panels, and poster presentations by leading international researchers, describing it as a platform designed to “spur dialogue and discussion on the next frontiers of AI research and its societal impact.”
Together, the opening addresses set the tone for a Symposium anchored in scientific rigor, global collaboration, and responsible innovation at a pivotal moment in AI’s evolution.
In his keynote session, Sir Demis Hassabis reflected on the rapid progress of AI since founding DeepMind in 2010, while cautioning that true artificial general intelligence (AGI) remains a work in progress. He outlined key technical gaps, including continual learning, long-term planning and consistency across tasks, even as he expressed strong optimism about AI ushering in a new era of scientific discovery. He closed with a message of “cautious optimism,” emphasising that while technical challenges can be solved through innovation, international cooperation will be essential to ensuring AI’s benefits are widely shared and its risks responsibly managed.
For the second keynote session, Dame Wendy Hall, Professor of Computer Science at the University of Southampton, examined the future of artificial intelligence through the lens of governance, inclusion and workforce transformation. Speaking in the context of the Global South, she called for AI systems that are built “for humanity”, grounded in safety frameworks, equitable access and sovereign capability. Emphasising that AI strategies must reflect national priorities, linguistic diversity and local data ecosystems, she highlighted the importance of inclusive development models that ensure long-term societal benefit. Her address conveyed cautious optimism, urging governments, researchers and young innovators, particularly in emerging economies, to shape AI in ways that are locally grounded, globally connected and fundamentally inclusive.
In his keynote address, Prof. Yoshua Bengio, Professor of Computer Science, Université de Montréal focused on the emerging risks associated with increasingly capable and agentic AI systems, particularly issues of misalignment, deceptive behaviour and inadequate risk mitigation. He warned that recent advances in AI capabilities are outpacing existing evaluation and safeguard mechanisms, highlighting concerns such as sycophancy, bias, jailbreaks, cyber misuse and systems exhibiting forms of self-preserving behaviour.
Emphasising that alignment challenges often arise as unintended side effects of current training methods, Prof. Bengio called for a fundamental shift in AI design, moving away from goal-driven, human-imitative systems toward models grounded in scientific reasoning.
Lastly, in his keynote address, Dr. Yann LeCun, Executive Chairman of AMI Labs and Professor of Computer Science at New York University, challenged prevailing narratives around artificial general intelligence, arguing that current AI systems, including large language models remain far from human-level intelligence. While acknowledging their impressive performance in language and narrow-domain tasks, he underscored their fundamental limitations: a lack of true understanding of the physical world, absence of persistent memory, weak long-term planning capabilities, and insufficiently robust safety controls.
To overcome these constraints, Dr. LeCun proposed a shift toward developing “world models”, predictive systems capable of simulating how environments evolve in response to actions. Such systems, he argued, would enable AI to anticipate consequences, plan effectively, and operate within defined safety guardrails, making them more controllable and aligned with human objectives.
As the Research Symposium on AI & Its Impact concluded, the discussions reflected both the extraordinary promise and profound responsibility that accompany rapid advances in artificial intelligence. From scientific discovery and global governance to alignment, safety and next-generation architectures, the Symposium underscored a shared imperative: to shape AI systems that are not only powerful, but trustworthy, inclusive and firmly aligned with human values.
