Supercharging AI Adoption in the Global South: Opportunities and Lessons from the Financial Sector

Detailed Summary

• John Tass Parker opened the session, positioning AI conversations in finance not around raw model performance but around legitimacy and trust. He argued that finance is moving from a “frontier‑AI” era to an institutional‑AI era where adoption hinges on auditability, resilience and regulatory supervision.
• He emphasised that model capability is increasingly commoditised; the differentiator is the ability of institutions to prove reliability to boards, regulators and customers.
• The moderator (named Bharat) introduced the panel and framed the discussion as a way to “capitalise the artificial‑intelligence movement for finance” and to explore how this can be exported to the Global South.

2. RBI’s Perspective – Enabling Responsible AI

2.1 Policy Stance

• Suvendu Parti (RBI) clarified that the RBI does not aim to “regulate AI” per se, but to enable responsible adoption.
• AI is seen as a strategic enabler of innovation while simultaneously demanding risk‑mitigation on bias, accountability, auditability and explainability.

2.2 Tech‑Neutral, Principle‑Based Regulation

• The RBI follows a technology‑agnostic approach, embedding AI concerns within existing consumer‑protection, IT‑outsourcing and concentration‑risk guidelines.
• The regulator highlighted a new AI‑specific report that distilled seven “sutras” (principles)—now adopted by the Indian government across sectors. One key sutra: “innovation over restraint” – encouraging firms to experiment responsibly.

2.3 Governance Expectations

• AI lifecycle management is expected to be a core consideration for banks, NBFCs and fintechs.
• Liability rests with the deployer (the regulated entity), not the model developer. The regulator called for:
  • Board‑level AI governance policies
  • Internal audit and assurance frameworks that capture incremental AI‑related risks (model drift, bias, hallucinations)
  • Transparency to customers – a “glass‑box” rather than a “black‑box” explanation, with an opt‑out option for non‑AI interactions.

2.4 Engagement Mechanisms

• RBI runs monthly “FinQuery / FinInteract” sessions (≈2,000 fintech participants over 1.5 years).
• A AI‑focused survey of >600 entities identified adoption gaps.
• The regulator operates a Regulatory Sandbox (since 2019) and is planning an AI‑Sandbox to provide affordable compute and data access for smaller fintechs.

3. JPMorgan Chase – Institutional Trust through Risk Management

• Terah Lyons described JPMorgan’s decade‑long journey from basic analytics to large‑language‑model (LLM) pilots and emerging agentic capabilities.
• Most impactful use cases identified:
  • Fraud detection & remediation – critical for safeguarding the sector.
  • Market‑risk analytics – speeding up decision making.
  • Compliance automation – ensuring regulatory adherence at scale.
• She praised the RBI’s principles‑based, tech‑neutral stance for allowing banks to experiment proportionately with risk.
• Lyons stressed that financial‑sector governance (model‑risk‑management, auditability, explainability) offers a template for other industries seeking responsible AI rollout.

4. Prosus Ventures – Investor View on AI’s Strategic Role

• Ashutosh Sharma framed AI as a strategic lever for three macro‑drivers in India’s fintech ecosystem:

  1. Unit‑economics – The Indian credit market (~$2trillion)spends3‑5$60‑100 bn annually). AI‑driven productivity can compress costs dramatically.

  2. Risk & “thin‑file” underwriting – A large share of the population lacks rich credit histories. AI can ingest unstructured data (e.g., mobile‑usage, social signals) to thicken credit files and expand access.

  3. Reach & Conversational Interfaces – Financial products are complex; a voice‑first, multilingual “app‑chat” can bring services to the billions who currently struggle with UI‑driven onboarding.

• Best‑practice highlights (still early):

  • Human‑in‑the‑loop – AI prepares the file; humans make the final decision.
  • Data‑privacy compliance – Strict adherence to India’s DPDP framework.

5. Razorpay – Agentic Commerce & Payments Innovation

• Harshil Mathur (referred to as “Harshal” in the transcript) described AI as a productivity multiplier for high‑volume data processing (risk, underwriting, fraud).

• Agentic commerce vision:

  • 300‑400 million Indian UPI users – only ~10 million drive 70 % of online commerce.
  • Existing apps are “American‑style supermarkets”; Indian consumers prefer conversation‑driven commerce (talking to a broker before buying).
  • Razorpay is building voice‑first, multilingual AI agents to bridge this gap, enabling conversational purchasing in sectors like travel, insurance and retail.

• Key technical constraints discussed later (data residency, hallucinations) also affect Razorpay’s rollout.

6. Distinguishing Regulators vs. Deployers

• RBI clarified its mandate limited to regulated entities (banks, NBFCs, fintechs). It does not directly regulate AI model developers (pure tech firms).
• The regulator emphasised a “glass‑box” approach: customers must be informed when interacting with AI and should be able to opt‑out.

7. JPMorgan Chase – Scaling Trust

• Terah Lyons reiterated that deployment without trust is impossible.

• She highlighted that risk‑management culture (model‑risk‑management guidelines, audit trails, transparency) is a competitive advantage enabling large banks to scale AI safely.

• A differentiated risk approach was advocated:

  • High‑risk applications (underwriting, payments) demand human‑in‑the‑loop and stringent validation.
  • Low‑risk services (document summarisation, onboarding assistance) can adopt lighter oversight.

8. Regulatory Challenges for Fintech Deployers

• Harshil Mathur identified three primary hurdles:

  1. Data‑residency – Indian regulations require customer data to stay within India. Many cutting‑edge LLMs are hosted abroad, limiting their use for PII‑sensitive workloads.

  2. Black‑box opacity – For LLMs, it is hard to trace data flow and guarantee that no unintended leakage occurs.

  3. Hallucinations – Even a 1 % error rate in financial advice can cause massive liability; current LLMs still produce inaccurate statements.

• He noted that domestic LLMs announced at the AI Summit could mitigate the residency issue, but model reliability remains a barrier.

9. Investor‑Side Regulatory Gaps

• Ashutosh Sharma praised RBI’s seven sutras as a useful north‑star but warned that regulators must keep pace with rapid AI evolution.

• He highlighted non‑regulatory bottlenecks that investors face:

  • Compute scarcity – Access to high‑performance GPUs/TPUs is limited for Indian startups.
  • Talent shortage – Research and engineering expertise are thin in the region.

• Sharma argued that policy alone cannot solve these constraints; public‑private partnerships (e.g., shared compute clouds, talent pipelines) are essential.

10. Future Bets – The Panel’s Vision for the Next Five Years

Speaker	Core Bet / Wish List
Suvendu Parti (RBI)	AI will dramatically expand financial inclusion through alternate‑data underwriting; AI‑driven assistive tech will bring un‑literate/disabled users into the formal banking system.
Harshil Mathur (Razorpay)	AI will lower servicing cost to the point where a village‑level user experiences a frictionless, voice‑first banking experience, eliminating the need for paper forms.
Terah Lyons (JPMorgan Chase)	AI will put a personal finance advisor in every pocket, democratizing wealth‑management that today only the affluent can afford.
Ashutosh Sharma (Prosus Ventures)	AI‑enabled multilingual, assistive banking will bridge the digital divide; language‑specific models will unlock mass adoption across India’s diverse linguistic landscape.
Collective	The sector must scale AI responsibly to avoid widening the digital divide, ensuring that trust mechanisms keep pace with capability.

11. Closing Remarks

• The moderator thanked the panel and reiterated that trust, legitimacy and inclusive design are the levers that will “supercharge AI adoption in the Global South.”
• Audience applause followed the final statements, marking the end of the session.

Key Takeaways

• Legitimacy over capability: In finance, the decisive factor for AI adoption is trustworthiness, auditability and regulatory compliance, not raw model performance.
• RBI’s tech‑neutral, principle‑based framework encourages innovation while mandating consumer‑protection, board‑level AI governance and a “glass‑box” transparency model.
• Financial institutions can export their AI‑governance playbooks to other sectors (e.g., agriculture, education) to accelerate responsible AI diffusion across the Global South.
• JPMorgan Chase’s risk‑management culture exemplifies how large banks can scale AI safely, focusing on high‑impact use cases such as fraud detection, compliance and market analytics.
• AI can thin‑file credit underwriting by leveraging unstructured data, unlocking credit for millions currently excluded from formal finance.
• Agentic, voice‑first commerce is seen as the next breakthrough for India’s massive but under‑digitised consumer base.
• Regulatory challenges for fintechs include data‑residency mandates, model opacity and hallucination risk; domestic LLMs and sandbox environments are early mitigations.
• Infrastructure gaps (compute, talent) are viewed by investors as the most pressing bottleneck, requiring coordinated public‑private interventions.
• A differentiated risk approach is recommended: high‑risk AI (underwriting, payments) warrants stringent human oversight; low‑risk AI (document summarisation) may adopt lighter controls.
• The panel’s collective vision: AI‑driven financial inclusion, multilingual conversational banking, and personal‑finance assistants will reshape the Indian and Global‑South financial landscape within the next five years.

---

See Also:

• the-governance-gap-designing-global-standards-for-ai-advisory-boards
• ai-for-economic-growth-and-social-good-ai-for-all-driving-economic-advancement-and-societal-well-being
• responsible-ai-from-principles-to-practice-in-corporate-india
• ai-innovators-exchange-accelerating-innovation-through-startup-and-industry-synergy
• democratizing-ai-resources-in-india
• effective-ai-assessments-verification-and-assurance-establishing-the-foundations-for-responsible-confidence-in-ai
• ai-and-indias-economic-growth-sectoral-impact-and-the-road-ahead
• ai-for-everyone-empowering-people-businesses-and-society
• ai-governance-in-the-age-of-powerful-ai-iinternational-perspectives-and-the-code-of-practice
• inclusion-for-social-empowerment