Scaling Trusted AI for 8 Billion+

Abstract

The panel explored how AI can be scaled responsibly to reach the world’s 8 billion + people while preserving trust. Dr. Aghi opened the conversation, framing AI as an inevitable, society‑wide “operating system” and warning against a future where a few entities control the technology. Brad Smith delivered a keynote that highlighted the growing North‑South AI divide, the need for infrastructure, skills, multilingual capability, and responsible governance. Panelists from Zoom, Rubrik, and Uniphore added perspectives on democratising AI through ease‑of‑use, entrepreneurship, application‑layer innovation, and data‑privacy safeguards. An extensive audience Q&A probed ROI, talent pipelines, safety, data‑ownership, and the role of India as an AI hub. The session closed with a visionary address by S. Krishnan, Secretary of the Ministry of Electronics & Information Technology, stressing India’s policy push for inclusive, voice‑first, multilingual AI ecosystems.

Detailed Summary

• Moderator welcomed attendees to the “US‑India Strategic Partnership Forum” panel on Scaling Trusted AI for 8 billion + people.
• Described AI as moving beyond “lab experiments” to become the “operating system of economies, institutions and societies.”
• Stressed that the real challenge is scaling trusted AI, not merely scaling AI itself.
• Emphasised that trust without access is meaningless and that the next billion innovators will emerge from small towns, farms, classrooms, and startups—not just Silicon Valley or Bengaluru.

2. Introductory Remarks – Dr. Mukesh Aghi

• Noted AI’s inevitability across every sector (education, health, transport, commerce).
• Framed the central question: Will AI create an equal, inclusive ecosystem or a new feudal system where few control the technology?
• Highlighted the panel composition (Microsoft, Zoom, Uniphore, Rubrik) and the need to discuss the direction AI will take, who controls it, and how benefits can be distributed broadly.

3. Keynote Address – Brad Smith (Microsoft)

Key Point	Details
AI Divide Statistics	By end‑2025, 25 % of the working‑age population in the Global North will be using AI vs. only 14 % in the Global South (a 25:14 gap). Growth rates: 1.8 % (North) vs. 1.0 % (South).
Historical Analogy	The divide mirrors the earlier electricity gap that powered industrialisation in the North; repeating it would be catastrophic.
Opportunity for the Global South	AI could be the greatest catch‑up engine of the 21st century if infrastructure is built now.
Infrastructure Pillars	Data centres, connectivity, electricity – all three must be delivered together. Microsoft plans to spend $70‑$78 bn in India and $50 bn globally in the Global South by 2030, though this is still insufficient.
Capital Mobilisation	Calls for a blend of private and public capital and demand‑generation to unlock further investment.
Skills Development	Skills are “the most important lever.” Microsoft aims to train 20 million people in India and launch 2 million educators via the Microsoft Elevate program.
Trust Enablers	1️⃣ Multilingual models – improve data, tools, and content provenance for all languages. 2️⃣ Application‑layer solutions – real‑world problems (e.g., food security, agricultural forecasting) must drive demand. 3️⃣ Responsible AI – companies can act responsibly today, irrespective of pending legislation.
Call to Partnership	Emphasised the core principle of the TRUST initiative: partnership across governments, industry, and academia.

Brad Smith concluded by reiterating Microsoft’s commitment to help close the AI divide and to work collaboratively under the TRUST partnership.

4. Panelist Opening Remarks

4.1 Aparna Bawa (Zoom) – “Democratise AI through Simplicity”

• Personal note: South‑Indian (Tamil) background, occasional language mix‑ups – symbolic of multilingual reality.
• Zoom’s “it just works” mantra ensures usability for anyone—from an 83‑year‑old grandmother to Fortune‑50 enterprises.
• AI is now seamlessly embedded in Zoom’s collaboration stack, providing transparent, “no hidden logic” experiences.
• Emphasised that natural‑language interfaces (e.g., ChatGPT) are the democratising force that lets non‑technical users harness AI.
• Stressed the need for “super‑simple, super‑easy, full‑scale” tools, especially for small‑and‑medium businesses and solopreneurs, to translate AI into productivity gains and national GDP growth.

4.2 Bipul Sinha (Rubrik) – “From Intuition to Innovation”

• Described the evolution from agrarian → industrial → knowledge → “intuition” economy.
• AI will automate knowledge, freeing humans for intuition work – connecting novel ideas that machines cannot yet infer.
• Highlighted language‑agnostic programming, allowing anyone to “talk to a computer” in their native tongue.
• Emphasised that AI removes knowledge barriers (e.g., a child in Patna has same information access as one in Palo Alto).
• Stressed the importance of data‑center tax incentives and the Indian government’s ₹150 bn investment to build the “factories of intelligence.”

4.3 Umesh Sachdev (Uniphore) – “Application‑First, Not Infrastructure‑First”

• Traced Uniphore’s origins to an IIT Madras incubator (2008), showing an Indian‑born AI company now serving 2,500 global customers.
• Asserted that India is not a runner‑up; the US and China dominate hardware & LLM development, but India can lead in AI diffusion, adoption, and applications.
• Warned against over‑regulation that could stifle innovation; advocated for a reactive regulatory approach matching AI’s rapid pace.
• Posited a “thousand‑flower” vision: for each sovereign LLM (SARVAM) there should be 5,000 AI‑application firms, producing a robust ecosystem that places India among the top three AI nations.

5. Audience‑Driven Q&A

Format – Moderator invited audience members to raise hands; each question was directed to one or more panelists. Below is a synthesis of the major queries and responses.

Questioner (Role)	Theme	Key Responses
Vijay (Cuts International – policy research)	ROI for large‑language‑model (LLM) investments in the Global South	• Rubrik’s Bipul acknowledged board pressure for ROI but warned that early‑stage tech adoption cannot wait for clear ROI; hesitation risks being left behind.
Dr. Aghi (follow‑up)	“Land‑grab” vs. trust – who will control hardware & data?	• Emphasised that few firms controlling hardware and datasets creates a new feudal structure; India should focus on application‑layer innovation and skilling to avoid becoming dependent.
Audience (unidentified)	Cheaper AI / data acquisition for developers	• Open call for technologists to reduce AI compute costs; highlighted need for affordable, open‑source tooling.
Student (final‑year BITS Pilani)	Advice for young graduates entering AI	• Bipul urged entrepreneurship over corporate jobs, citing personal experience of early investment in OpenAI; emphasised risk‑taking and building homegrown products.
Sejal (AWE Funds)	Safety, privacy, and recent “Claude safety officer” resignation	• Rubrik’s Bipul explained Rubrik’s guard‑rail framework: fine‑tuning models, halting unsafe actions, and providing deterministic outputs to preserve trust.
Unnamed audience	Equitable sharing of data‑derived profits for Global South contributors	• Rubrik’s stance: does not train its models on customer data; positions itself as a trusted partner that respects data ownership.
Vijay (again)	Whether India can realistically compete with US/China	• Dr. Aghi argued that India’s strength lies in its talent pool; by 2030, 20 % of the world’s digital workers could be Indian. Focus on app‑side rather than hardware.
Maria (policy perspective)	Funding gap for Indian AI startups	• Dr. Aghi noted that ~90 % of current AI funding comes from US sources; US‑India Strategic Partnership Forum is expanding a startup program (legal, financing, market access) to bridge the gap.
Audience – data‑ownership	Who profits from the data used to train LLMs?	• Rubrik reaffirmed its no‑training‑on‑customer‑data policy, arguing that ethical data use is a competitive differentiator.
S. Krishnan (Ministerial Secretary – closing remarks)	Policy roadmap and practical steps for inclusive AI	• Stressed voice‑first, multilingual AI (via Bhashini & other initiatives). • Mentioned AI‑CORe (Compute‑On‑Request) enabling Indian compute cost at ≈ ⅓ global price and a 7,500‑dataset repository (growing rapidly). •强调 application‑centric models (small, domain‑specific) over massive LLMs for agriculture, health, education, manufacturing.
Additional audience queries (various)	Topics ranged from AI‑driven job creation, sovereign AI models (“SARVAM”), collaboration across fragmented initiatives, and the role of Indian diaspora in bridging ecosystems.	• Consensus: collaboration, open data, and rapid skill‑up are essential; regulatory balance needed to avoid stifling innovation.

6. Closing Remarks – S. Krishnan (Secretary, Ministry of Electronics & Information Technology)

• Congratulated organisers and participants, underlining the democratic intent behind the summit.
• Highlighted India’s mobile‑first, voice‑first reality; AI must be multilingual (22 official languages + dialects).
• Described the AI‑CORe initiative that reduces compute costs and expands public datasets (from 300 to 7,500 in six months).
• Stressed the importance of application‑layer solutions – small, domain‑specific models for agriculture, health, education, manufacturing.
• Reiterated that India’s AI journey mirrors the Y2K moment, requiring collective, open‑minded innovation.
• Expressed optimism: AI will create jobs, not eliminate them, and will help India become a “Viksit Bharat” (developed nation) by 2047.
• Invited attendees to explore the AI Expo (≈ 1,000 startups) and thanked everyone for their energy despite logistical challenges (weather, traffic).

Key Takeaways

• The AI Divide is Real & Widening – By 2025 only 14 % of the Global South’s workforce will be using AI versus 25 % in the North; urgent infrastructure and skill interventions are required.
• Infrastructure + Skills = Foundation – Data centres, reliable electricity, and broadband must be built alongside massive up‑skilling programmes (e.g., Microsoft’s 20 M‑person target in India).
• Multilingual, Voice‑First AI is Critical for India – With 22+ official languages, AI must understand and generate in native tongues; initiatives like Bhashini and AI‑CORe are tackling data‑availability and cost barriers.
• Trust Requires Three Pillars: (1) Multilingual model performance, (2) Application‑layer relevance (real‑world problems like food security), (3) Responsible governance (guard‑rails, privacy, data‑ownership).
• India’s Competitive Edge Lies in Applications, Not Hardware – While the US and China dominate LLM and GPU development, India can lead in AI diffusion, adoption, and domain‑specific solutions.
• Entrepreneurship Over Employment – Panelists urged graduates to become job creators; early‑stage risk‑taking can yield outsized returns (e.g., early OpenAI investors).
• Policy Must Be Enabling, Not Stifling – Over‑regulation can choke innovation; a reactive approach adapted to AI’s speed is preferred.
• Data Ownership Matters – Rubrik’s policy of not using customer data for model training showcases a trust‑building business model.
• Public‑Private Partnerships are Essential – US‑India Strategic Partnership Forum’s startup program and Microsoft’s capital commitments illustrate the need for joint financing, market access, and legal support.
• AI Will Create Jobs, Not Destroy Them – Both Uniphore and Microsoft foresee an abundance era where AI‑driven productivity fuels new employment opportunities, especially in application development and domain expertise.

---

The session offered a comprehensive view of the challenges and opportunities inherent in scaling trusted AI for the entire global population, marrying high‑level policy insight with on‑the‑ground entrepreneurial perspectives.

See Also:

• responsible-ai-at-scale-governance-integrity-and-cyber-readiness-for-a-changing-world
• ai-innovators-exchange-accelerating-innovation-through-startup-and-industry-synergy
• responsible-ai-in-action-how-global-enterprises-are-building-trust-at-scale
• ai-for-everyone-empowering-people-businesses-and-society
• trusted-ai-practical-pathways-for-the-public-sector
• safe-ai-building-shared-trust-and-accountability-infrastructure
• aligning-ai-governance-across-the-technology-stack
• building-a-trusted-and-resilient-ai-infrastructure-ecosystem-balancing-innovation-security-and-rights
• shaping-resilient-futures-in-the-age-of-ai-leadership-for-the-technology-energy-and-security-transitions