India’s AI Leap: Policy to Practice with AIP2

Contents

Executive Summary

This panel discussion from an AI summit in India emphasizes that AI diffusion in the Global South must move beyond technological capability to focus on inclusive, equitable deployment through infrastructure, skills development, and trust-building standards. Speakers stress that effective AI adoption requires simultaneous investment in digital connectivity, workforce upskilling, governance frameworks, and public literacy—particularly in institutionally fragile contexts—rather than pursuing "moonshots" disconnected from real-world human needs.

Key Takeaways

AI diffusion ≠ technology deployment alone: Genuine diffusion requires simultaneous action on five dimensions: infrastructure, data/trust, institutions, skills, and market shaping. Without addressing connectivity, literacy, and institutional capacity together, technology access will exacerbate inequalities.
Build trust through clear governance, not ethical appeals: Vague commitments to "AI ethics" are insufficient, especially in fragile institutional contexts. Governance must specify what is prohibited (e.g., social scoring, emotion recognition in workplaces), what is regulated with standards (e.g., training data adequacy, cybersecurity thresholds), and what is left to market forces.
Standards-setting needs acceleration and accountability: Standards are being delayed by actors resistant to compliance. Time-bound commitments and mechanisms to prevent indefinite delays are needed, or key governance implementations will never materialize.
Participatory governance means meaningful inclusion, not tokenism: Representation from Africa, Latin America, and Asia in standards committees, standards co-authorship, and leadership positions requires dedicated funding and long-term commitment. Past processes excluded those with fewer resources.
Skills investment is the highest-return use of capital: From basic literacy in schools to policy-maker training to civil society capacity-building, closing the AI understanding gap is prerequisite to all other diffusion efforts and the most difficult to shortcut.

Key Topics Covered

AI Infrastructure & Connectivity: Digital public infrastructure as a foundation for AI accessibility; connecting offline populations globally
Skills & Literacy Gaps: Critical importance of upskilling programs, digital agency, and public understanding of AI across all demographic groups
Standards & Governance: Role of AI standards in ensuring interoperability, trust, and mitigation of risks (deepfakes, misinformation, surveillance)
Global South AI Diffusion: Context-specific approaches vs. one-size-fits-all regulatory models; participatory governance frameworks
Startup Ecosystems & Innovation Transfer: How startups bridge technology-business gaps and enable corporate transformation
Labor Displacement & Democratic Institutions: Risks of AI-enabled job displacement and surveillance; need for strong institutional capacity
Public Perception & Democratic Participation: Wide knowledge gaps in public understanding of AI limiting citizen participation in governance decisions
EU AI Act Implementation: Practical regulatory lessons; distinction between prohibitions, high-risk regulations, and light-touch areas
Dual-Use Technology Risks: How beneficial AI tools (e.g., voice-based diabetes detection) can enable surveillance if unchecked
Regional Governance Approaches: Comparative analysis of EU (rights-focused regulation), US (innovation-focused), China (state-coordinated), and Global South (developmental outcomes-focused) strategies

Key Points & Insights

Connectivity is a prerequisite, not a consequence: One-third of humanity remains offline. Without foundational digital infrastructure (the "giga initiative" targeting school connectivity and "hardest to connect" communities), AI diffusion cannot reach those who need it most. Current commitments stand at ~80 billion of a 100 billion target.
Skills development as "the engine of agency": Connectivity alone creates digital access; skills development is what enables meaningful participation and agency. India's Future Skills program and ITU's skilling coalition (70+ partners, 180 learning resources in 13 languages) offer scalable models, but the global AI skills gap remains massive.
Public literacy is dangerously low in Global South contexts: Survey data from South Africa showed two-thirds of the population lack meaningful AI understanding. Large-scale AI adoption is outpacing public awareness, creating a "democratic gap"—people cannot contest or participate in decisions affecting them. This is especially critical where AI is deployed in public services.
Standards are governance made concrete: Voluntary standards developed through inclusive multi-stakeholder processes are not optional luxuries—they are essential for embedding trust, ensuring interoperability, and combating deepfakes/misinformation. However, standard-setting is being deliberately delayed by some private sector actors resistant to compliance requirements.
One-size-fits-all regulation risks reproducing dominance: Global North regulatory frameworks (GDPR, EU AI Act) have had limiting effects on developing regions. Instead, global consensus should focus on shared principles (accountability, transparency, safety, human oversight) while allowing regional adaptation based on local context, capacity, and institutions.
Institutional capacity is a prerequisite for rights protection: Democracy commissions, human rights bodies, competition regulators, and information commissioners in Global South countries must be strengthened to counterbalance big tech monopolies and prevent AI-enabled mass surveillance and repression—especially in institutionally fragile contexts.
The "technology overshoot" problem in SMEs: Small and medium enterprises in many countries face a gap where AI capability has advanced faster than their understanding of business integration needs. Startups are uniquely positioned to bridge this gap because they understand both technology and business transformation.
Dual-use dilemma requires proactive design: Voice-based diabetes detection can be beneficial, but the same technology can reveal sleep patterns, diet, medication use, and attention levels. Beneficial applications cannot be taken for granted; ethics, security, human rights, and sustainability must be "baked in" from inception, not added retrospectively.
India's model demonstrates AI diffusion at scale: Digital ID (Aadhaar), financial inclusion systems, and multilingual digital public infrastructure (Bhashini in 22 languages) serving 1 billion+ people provides a proof-of-concept for equitable AI deployment in fragile institutional contexts. This is distinct from typical "moonshot" narratives.
Startups as "transmission mechanisms" for policy to impact: Over 6,000+ startups in India's ecosystem, with ~1,000 crores in direct funding plus 8,000 crores from the India AI Mission, show that capital scarcity is not the limiting factor. The constraint is matching startups with market access, mentorship, and clear policy direction.

Notable Quotes or Statements

Dorin (ITU): "AI diffusion isn't about everyone using the same technology. It's about giving everyone the same bridge to opportunity and refusing to let the digital divide become an AI divide."
Dorin (ITU): "Without connectivity, there is no AI."
Dr. Madan Gopal (Matey Startup Hub): "Your customer is your best investor if you're a startup... finding customers for startups is more important than finding investors."
Rachel Adam (Global Centre on AI Governance): "People don't know about these technologies. They don't know about the risks. They don't know about the opportunities. They're not able to contest it. They're not able to participate in decision making. We have a real problem."
Brando Benzi (EU Parliament, AI Act co-drafter): "Enthusiasm for diffusion should not be a substitution for building frameworks that are precise and not generic ethical appeals... if you substitute regulation, governance of all kinds with merely voluntary ethical frameworks, I'm not sure we are getting anywhere."
Fred Werner (ITU): "How do you take these ambitious words and text and turn them from principles to implementation? Because the devil is in the details and standards have details."
Rachel Adam: "We want to be seeing kind of a global consensus around a set of principles... but noting that different regions are going to need to adapt those standards in different ways."

Speakers & Organizations Mentioned

Speaker/Role	Organization	Affiliation
Dorin (keynote speaker)	ITU (International Telecommunication Union)	Chief of strategy for AI for good
Fred Werner	ITU	Chief of Strategic Engagement; co-creator of AI for Good Global Summit
Dr. Rachel Adam	Global Centre on AI Governance	Founder & CEO; advisor to governments; contributor to African Union AI strategy
Brando Benzi	European Parliament	Co-drafter of EU AI Act; Italian MEP since 2014
Dr. Madan Gopal	Matey Startup Hub	CEO; custodian of deep tech startups in India
Prime Minister Modi	Government of India	Referenced for commitment to human-centered AI
—	UNICEF	Partner on "giga initiative" for school connectivity
—	ISO/IEC	Partners with ITU on AI standards development
—	Ministry of External Affairs (India)	Co-organizer of summit with Matey
—	Innovation Factory	AI startup hub running year-round acceleration
—	C2PA (Coalition for Content Provenance and Authenticity)	Partner on multimedia authenticity standards

Technical Concepts & Resources

Bhashini Platform: Multilingual digital public infrastructure delivering government services in 22 languages; example of accessible AI-powered infrastructure in India
AI Standards Exchange Database: Launched by ITU with 850+ standards and technical publications; includes multimedia authenticity standards for deepfake detection/traceability
C2PA Standards: Multimedia content provenance and authenticity framework being developed for deepfake/misinformation detection
Aadhaar (Digital ID): India's biometric identity system serving 1 billion+ people; referenced as proof-of-concept for large-scale equitable technology diffusion
EU AI Act: World's first comprehensive AI regulation; distinguishes between:
- Prohibited use cases: social scoring, predictive policing, emotion recognition in workplaces/study places, manipulative/subliminal techniques
- High-risk applications: requiring standards on data adequacy and cybersecurity thresholds
- Light-touch areas: non-regulated beyond existing legislation
Global South AI Diffusion Playbook: Framework with five interacting dimensions: infrastructure, data/trust, institutions for procurement/market, skills, and market shaping
AI Skills Coalition (ITU): 70+ partners, 180 learning resources across 13 languages
India AI Mission: ~8,000 crores committed for startup funding
Giga Initiative: UNICEF partnership to connect every school globally; current target ~100 billion USD in commitments/pledges
Voice-Based Health Detection (Estonian startup example): AI application detecting blood sugar levels from voice patterns via mobile phone; illustrates dual-use potential (beneficial and surveillance-enabling)
AI for Good Global Summit: Held July 7-10 in Geneva; annual hub for collaboration on standards and AI-driven impact
African Union Continental AI Strategy: Examples of Global South governance approaches emphasizing human rights, gender, and children's rights

Document Quality Note: The transcript contains some audio artifacts and incomplete sentences reflective of live speech, but all substantive claims and arguments have been preserved accurately as they appear in the source material.