
P E D A G O G Y F O R T H E 2 1 S T C E N T U R Y
11
Foreword | Joining Plaksha | Freshmore Curriculum | Selecting a Major | Electives & Minors| Research Mindset |
Entrepreneurship Mindset | Beyond the Classroom | Beyond Plaksha
If AI researchers in India are to create lasting
research impact, it must be grounded in the
realities it seeks to serve. A central insight that
emerged repeatedly at ConfAI 2025 was the
urgent need to focus on creating and using
Indian datasets. In most real-world
applications, datasets developed in the US and
Europe (and made publicly available) have
limited relevance for Indian contexts.
This was one of the key takeaways at ConfAI
2025, held successfully November 28-30 at the
Plaksha campus in Mohali, under the theme
'Conversations that shape AI futures'. This 3-
day academic conference brought together 11
eminent speakers from academia, industry
and government, alongside high-quality
student presentations and posters based on
work accepted at leading international AI
conferences.
Several talks reinforced the importance of
trust, fairness and reasoning as core
challenges in AI systems. Dr Venkat
Padmanabhan (Microsoft Research India (MRI))
set the tone by going beyond AI to define what
would constitute impactful research. He
presented examples from decades of work at
MRI. Dr Mausam (IIT Delhi) focused on
reasoning, while Dr Richa Singh (IIT Jodhpur)
highlighted examples of bias in AI inference.
Dr Sunita Sarawagi (IIT Bombay) discussed
advances in structured learning, and Dr Mayur
Naik (UPenn) introduced neuro-symbolic AI as
a promising pathway to support reasoning in
AI tools.
The second major insight was on small,
application-specific models that can be
deployed on edge devices - mobiles,
wearables, and low-cost hardware, rather than
large, resource-intensive systems. These
models can be trained on Indian datasets and
made capable of communicating in Indian
languages. Such systems can create impact in
research as well as startup-led innovation.
Concerns around efficiency and sustainability
surfaced across multiple sessions. Dr Mayank
Vatsa (IIT Jodhpur) talked about efficient
unlearning, while Dr Niket Tandon (MRI)
addressed memory efficiency. Dr CV Jawahar
(IIIT Hyderabad) advocated for the creation of
Indian datasets and associated solutions, while
Dr Renu Rameshan (Vehant Technologies)
described the practical challenges of deploying
AI systems in the field. Debjani Ghosh (NITI
Aayog) outlined ongoing government
initiatives and articulated her views on areas
where India must focus its efforts.
A third takeaway that emerged was on the
need to be extremely innovative if one is
pursuing mainstream AI research, particularly
Large Language Models (LLMs) as the resource
gap between academia and industry is
substantial. This theme surfaced during the
panel discussion 'Beyond the Hype: Core
Challenges in AI Research,' featuring Dr
Mausam, Dr Sarawagi, and Dr Jawahar,
moderated by Dr Siddharth. The conversation
generated both light and heat.
The conference concluded with a reflective
keynote by Dr Rajesh Gupta (UC San Diego),
who traced the historical evolution of technical
education in the US and what needs to change
with AI tools emerging as key learning aids.
Equally encouraging was the quality of student
participation throughout ConfAI 2025. Student
talks were organized into sessions and
reinforcing the promise of a new generation of
engineers and researchers ready to engage
with complex societal challenges.
Shaping India's AI future will require not only
technical excellence, but also contextual
grounding, interdisciplinary thinking, and the
courage to pursue research agendas that
matter.
ConfAI 2025: Conversations that shape
AI future
The 3-day academic conference sparked critical conversations that highlighted
the need for Indian datasets, context-aware models, and innovative academic
research to create meaningful, real-world impact.
Prof M Balakrishnan
Distinguished Visiting Professor
Plaksha University
T H E U N D E R G R A D U AT E J O U R N E Y AT P L A K S H A