India AI Impact Summit Insights
All sessions

Scaling Trusted AI: Global Practices, Local Impact

Contents

Executive Summary

This talk, delivered at an India AI Impact Summit, argues that the next wave of AI development will be defined not by raw capability but by measurable trust and responsible governance. The speaker emphasizes that India's unique position—with 1.4 billion citizens, digital infrastructure (Aadhaar, UPI), and government-led AI governance frameworks—positions it to lead the global south in building trustworthy, context-aware AI systems that serve populations often overlooked by Western AI development.

Key Takeaways

  1. Trust is the new competitive moat: Enterprises and governments increasingly select vendors and partners based on measurable trustworthiness, not just capabilities. Building trust evidence into your AI systems is market access, not compliance overhead.

  2. Governance must be full-lifecycle and contextualized: Embed responsible AI from data collection through deployment and monitoring. Tailor frameworks to your industry, region, and specific risk profile—not abstract global standards.

  3. India is uniquely positioned to lead global south AI: Billion-user digital infrastructure (Aadhaar, UPI), government governance frameworks, and cultural/linguistic diversity give India template potential for trustworthy AI serving underserved populations worldwide.

  4. AI governance is a human-skills growth engine: New professional roles in AI evaluation, governance, auditing, and agentic system management require irreducibly human judgment and ethical reasoning—expect substantial, immediate job creation in these areas.

  5. Responsible AI accelerates innovation, not inhibits it: When governance provides clarity and confidence from day one, it reduces risk, enables faster iteration, builds customer trust, and unlocks market access—the opposite of friction.

Key Topics Covered

  • AI's Economic Impact & Current Adoption in India

    • AI projected to contribute $500 billion to Indian GDP by 2030
    • 47% of Indian enterprises already have multiple AI use cases in production (not pilots)
    • Acceleration across precision agriculture, fraud detection, rural healthcare diagnostics

  • Trust as Competitive Advantage, Not Regulatory Burden

    • Governance framed as a market access requirement and strategic advantage
    • Enterprises, governments, and trading partners increasingly demand trustworthiness evidence
    • Trust embedded in standards, institutions, and procurement processes

  • India's Government-Led AI Governance Framework

    • AI governance guidelines released with seven actionable "sutras"
    • "Technolegal" framework for AI governance (compliance by design, not afterthought)
    • Framework spans full AI lifecycle: data collection → model training → deployment → autonomous agents

  • Context-Specific AI Governance

    • One-size-fits-all ("peanut butter") approach to AI governance inadequate
    • Governance must account for industry, region, risk profile, use case
    • Same AI system can be helpful or harmful depending on deployment context

  • AI as Job Creator (Emerging Professional Roles)

    • New professional categories emerging: AI evaluators, AI governors, autonomous agent managers, trust & verification specialists
    • These roles require distinctly human skills: judgment, contextual reasoning, ethical analysis, stakeholder communication

  • Global AI Governance Hub & Resource

    • Launch of first global AI governance insights hub (free for global south)
    • Single source of truth aggregating: policies, risk taxonomies, control frameworks across 72+ countries
    • Community-driven model with voting on risk mappings, policy impact ratings, severity assessments

  • Real-World Implementation Across Enterprise Sectors

    • Mastercard: AI for fraud detection (300% improvement in speed/accuracy); agentic AI now available in India
    • PepsiCo: Responsible AI as front-office function; integration across farm-to-consumer value chain; WhatsApp ordering via AI; governance embedded in SDLC
    • PB Fintech: Handling ~250 million customers/year; AI for insurance/lending decisions; focus on financial literacy and inclusivity for 140 crore population; healthcare expansion (PB Health)
    • G42: Building sovereign, multilingual frontier models (Hindi, English, Arabic, Kazakh); infrastructure at national/regional scale; digital schools pilot in Azerbaijan

  • Agentic AI Governance Challenges

    • Autonomous agents introduce new governance complexities
    • Mastercard framing agents as equally safe/secure transactions; building standards for trustworthy agentic systems

  • Data, Cybersecurity & Cross-Border Cooperation

    • AI lowers barriers to cyber crime (deepfakes, social engineering)
    • Need for shared intelligence, data-sharing across borders and public-private sectors
    • Standard fragmentation and data silos benefit cyber criminals

Key Points & Insights

  1. Trust is the unlock: The next competitive advantage in AI isn't speed or model capability—it's measurable, provable trust. Organizations answering "yes, with evidence" to trustworthiness questions win contracts and market access.

  2. Governance is not an afterthought: The most successful organizations embed responsible AI throughout the full AI lifecycle (from data collection through deployment and monitoring), not as an airport security checkpoint before launch.

  3. Context is everything: The same AI system can be helpful or harmful depending on where and how it is deployed. Governance frameworks must be tailored to industry, region, cultural factors, and specific risk profiles—not abstract or universal.

  4. India's structural advantages: India's experience building billion-user systems (Aadhaar, UPI) with institutional trust, combined with government-led governance frameworks and digital infrastructure diversity, positions it uniquely to lead responsible AI for the global south.

  5. New job categories require human judgment: AI governance is creating substantial, immediate employment (not speculative future roles) in areas requiring irreducibly human skills: contextual reasoning, ethical analysis, stakeholder communication, stress-testing, auditing.

  6. Governance as strategic, not compliance burden: Organizations framing responsible AI as a business advantage (clarity, confidence, market access, brand protection) outpace those treating it as regulatory friction.

  7. Multilingual, multicultural frontier models essential: Frontier AI developed only in English/US context fails in global south contexts. G42's work on Hindi, Arabic, Kazakh models demonstrates necessity of regional AI infrastructure.

  8. Shared intelligence architecture critical: Cyber threats from AI (deepfakes, social engineering) amplified; combating them requires cross-border, public-private data-sharing and unified standards—not fragmented regional approaches.

  9. Enterprise examples validate feasibility: Mastercard (fintech), PepsiCo (consumer goods), PB Fintech (BFSI/health), and G42 (frontier models) demonstrate responsible AI is operationalizable at scale across diverse sectors.

  10. Inclusive financial/health AI is both ethical and viable: PB Fintech's focus on 140 crore population, with emphasis on middle/lower-middle class, financial literacy, and inclusive health insurance, shows responsible AI directly addresses development priorities.

Notable Quotes or Statements

"The next wave of AI will not be defined by capability alone. It will be defined by trust that you can define, measure and prove." — Speaker (Navina/Credo AI executive)

"India is not just asking the question should we build AI? The exciting thing is India is asking the question which is a hard one: How do we build AI that works for 1.4 billion people and how do we prove it can be trusted?" — Speaker, on India's governance approach

"Responsible AI is not an airport security check—it's really embedded as a strategic imperative in everything we do." — Fabris Cespedes, G42

"For us, the only AI is responsible AI. It is all about trust—way beyond regulations." — Caroline Levu, Mastercard

"Responsible AI is not a back office function...it's really a front office function because it truly is welfare for all." — Magesh Bwati, PepsiCo

"Governance is becoming a market access requirement. Enterprises want to know can I trust the AI? Governments want to know does it meet our standards. Trading partners want to know can I rely on this AI across borders?" — Speaker, framing governance as competitive advantage

"AI governance is going to be the next frontier of job creation that humans are going to be critical for." — Speaker, on emerging professional roles

"We must join forces. We must act together and we must be able to share data intelligence across borders and across the public and private sector. That is absolutely mission critical." — Caroline Levu, on cyber threats and cross-border cooperation

"If we don't adopt AI, we don't get enough data points to improve AI and we don't make AI safer and more trusted." — Fabris Cespedes, G42, on adoption as prerequisite for safety

Speakers & Organizations Mentioned

Primary Speaker:

  • Navina Ekka (or similar name) - CEO/Executive, Credo AI; specialist in AI governance

Panel Speakers:

  • Fabris Cespedes (or Frisse/Frix) - Group Head of Responsible AI, G42 (UAE-based technology company)
  • Caroline Levu - Chief Privacy, AI and Data Responsibility Officer, Mastercard
  • Magesh Bwati - Senior Vice President and Global Head of Data, Analytics & AI, PepsiCo
  • Rajiv Gupta (Rajie) - President, PB Fintech (PolicyBazaar + PesaBazaar + PB Health)

Organizations/Entities Referenced:

  • Credo AI (AI governance software/platform provider)
  • G42 (Abu Dhabi-based AI infrastructure, data centers, frontier model developer)
  • Mastercard (fintech, fraud detection, agentic AI)
  • PepsiCo (consumer goods, FMCG; brands: Mountain Dew, Lays, Doritos, Kurkure)
  • PB Fintech (PolicyBazaar, PesaBazaar lending platform, PB Health)
  • Meta (partnership on WhatsApp ordering)
  • Indian Government Office of Principal Scientific Adviser to PM Modi (technolegal AI governance framework)
  • Various regulators: RBI (Reserve Bank of India), IRDA (insurance regulator), EU, SEC

Technical Concepts & Resources

AI Governance Frameworks & Standards:

  • Seven Sutras - India's AI governance guidelines (actionable pathways for builders/enterprises/institutions)
  • Technolegal Framework - Governance embedded throughout AI lifecycle, measured scientifically; compliance by design
  • EU AI Act - Referenced regulatory comparison
  • 72 countries with 1,000+ policy initiatives - Current global fragmentation of AI policy

Risk & Governance Taxonomy:

  • 16 distinct risk types mapped: policy violations, bias, security vulnerabilities, hallucinations
  • Risk control mappings and mitigation strategies (human-in-loop oversight, technical stress tests)
  • Policy tracker, risk taxonomy, risk/controls framework (three pillars of Credo AI's hub)

AI Applications Mentioned:

  • Precision agriculture
  • Fraud detection (banking/fintech)
  • AI-powered diagnostics (rural healthcare)
  • Facial recognition systems
  • Large language models (LLMs)
  • Agentic AI / autonomous agents
  • Deepfakes & social engineering threats
  • Multilingual models (Hindi, English, Arabic, Kazakh)
  • Generative AI techniques (300% improvement in fraud detection speed/accuracy at Mastercard)

Tools/Platforms:

  • Credo AI's Global AI Governance Insights Hub - Aggregates policies, risk controls, vendor assessments across 72+ countries (free for global south)
  • Credo AI's AI Governance Software - Operationalizes governance; daily workflows, measurable checks, organizational alignment
  • Mastercard AI Garage (Pune, India) - R&D hub with 6,000 technologists; research on global south applications

Infrastructure/Systems Referenced:

  • Aadhaar - Digital identity system (1.3 billion enrollments)
  • UPI - Unified Payments Interface (13+ billion transactions/month)
  • WhatsApp ordering integration - AI-powered retail ordering

Professional Roles (Emerging):

  • AI evaluators
  • AI governors
  • Autonomous agent managers
  • Trust & verification specialists

Policy & Governance Implications

  • India's government-led approach to AI governance (rather than purely regulatory/reactive) positions it as potential template for global south
  • Purpose-built institutions tailored to India's sectoral diversity and digital infrastructure emphasized as model
  • Cross-border standard-setting and data-sharing critical to combating cyber threats amplified by AI
  • Procurement as enforcement mechanism: Organizations incentivized to embed trust via vendor selection, contract requirements
  • Community-driven governance intelligence: Credo AI hub enables crowdsourced validation of policies, risks, and controls

Production Quality Note: The transcript contains repetition artifacts (words/phrases repeated 2-3 times consecutively), suggesting OCR or transcription processing errors. Core content remains clear and substantive.