AI Impact Summit Day 4: Amodei and Pichai warn the same thing: Don’t let AI widen the divide

The AI Impact Summit Day 4 opened with the “build” side of the story from Ashwini Vaishnaw, Union Minister for Railways; Information & Broadcasting; and Electronics & Information Technology, Government of India. He spoke about compute as shared national infrastructure, not a luxury product. At the centre of that was a common compute platform built through public-private partnership, with access to 38,000 GPUs for startups, academia, researchers and students, and 20,000 more planned.

That access theme ran through rest of the speeches as well. 

Amodei: manage the disruption, don’t deny it

It came through most sharply in Dario Amodei, CEO, Anthropic, who stayed focused on what happens when AI moves from novelty to daily work. His belief was optimistic. AI can grow the economic pie, including in India and the Global South. But he kept coming back to the speed of change. When disruption arrives faster than adaptation, the winners run ahead and everyone else gets left negotiating the fallout.

Amodei’s remedy was plain: don’t treat the economic impact as a later problem. Bring economists, labour leaders, policymakers, and companies into the same room early, and keep them there. Use real-world signals to understand what is changing in jobs, tasks and skills, then adjust faster than governments typically do. He spoke about managing what he called the “time of disruption” so the path from change to prosperity is not chaotic and cruel, but smoother and more widely shared. Anthropic has been building public work around this through its Economic Futures programme and the Economic Index, to track real usage patterns and connect them to policy conversations.

Pichai: bold AI needs infrastructure, inclusion and trust

Then Sundar Pichai, CEO, Google and Alphabet, took the stage and talked about the same divide- warning and his speech anchored in infrastructure and lived experience. He spoke about passing through Visakhapatnam as a student on the Coromandel Express and remembering it as a quiet coastal city with promise. Now, he said, it is becoming a major AI hub in Google’s India plans: a full-stack AI hub in Vizag, part of Google’s $15 billion infrastructure investment, with gigawatt-scale compute and a new international subsea cable gateway intended to spread jobs and capability across the country.

Pichai did not present this as glamour spending. He presented it as the difference between inclusion and exclusion. If AI is going to reshape education, health, farming, and enterprise productivity, then compute and connectivity cannot stay concentrated. That is why he tied the Vizag hub to America India Connect, Google’s connectivity push anchored in the same five-year India investment, aimed at strengthening digital reach and resilience across regions.

He also worked hard to show what “bold” AI looks like when it actually serves people. He pointed to big science, especially AlphaFold, which helped crack the protein structure prediction problem and was recognised in the Nobel Prize in Chemistry 2024 for work that included AI-based protein structure prediction by Demis Hassabis and John Jumper. The point was not the trophy. It was the acceleration: decades of discovery compressed into something researchers around the world can build on.

But he kept moving back to everyday outcomes. Health access for people who cannot reach doctors. Better forecasting for farmers facing unpredictable monsoons. Tools that work across languages. In his telling, AI’s promise is real, but not automatic. Emerging economies can leapfrog, yes, but only if the basics are built and the benefits do not get trapped inside a few companies, a few cities, and a few languages.

Pichai also returned to a hard truth that sits behind both his message and Amodei’s: trust is not optional. As synthetic content gets easier to produce, adoption breaks down when people stop believing what they see and read. He pointed to verification tools such as SynthID, used by journalists and fact-checkers, as one part of keeping public trust intact in the AI era.

PM Modi’s address: practical outcomes, not abstract AI

Prime Minister Narendra Modi kept his message direct and grounded. He called AI a turning point in human history, but quickly brought it back to what people can actually use. He pointed to the energy he saw among young builders at the summit and said India’s youth is not just learning AI, but taking ownership of it.

Also read: Prime Minister Narendra Modi’s AI Impact Summit inaugural address

He linked that confidence to real needs: better support for agriculture, stronger security use cases, practical help for persons with disabilities, and tools that work across India’s multilingual population. The emphasis was not on abstract capability. It was on AI that delivers in the field, in local languages, and at a scale that matches India. He used that as a signal of what “Made in India” can look like in the AI era: not just building models, but building solutions people can feel.

The outcome moment: frontier AI commitments

What made Day 4 start important was that the biggest speakers were not only selling possibility. They were also talking about the costs, jobs, trust, and inclusion, without pretending these will solve themselves.

That is why the summit’s “outcome moment” mattered: the announcement of the New Delhi Frontier AI Impact commitments. In the transcript, the themes were simple and grounded: share anonymised and aggregated insights to support evidence-based policymaking on jobs and skills, and strengthen multilingual, contextual evaluations so AI systems are tested across languages, cultures, and real-world use, especially in Global South settings. It matched what Amodei and Pichai were both signalling in different ways. If you do not measure what is changing, you cannot manage it. If systems do not work across language and context, “AI for all” stays a slogan.

The trajectory Day 4 pointed to

Meanwhile, with leaders likeAntónio Guterres, Secretary-General of the United Nations, and Emmanuel Macron, President of France, also pushing inclusion, sustainability and safety, the theme was already set.

Day 4 has started with the right intentions, but the real test sits ahead. It will come down to who builds the rails, who shares access, who protects trust, and who takes responsibility for the disruption that will come alongside the gains.

The AI heavyweights on Day 4 did not just deliver speeches. They set the tone for where this is headed. Put together, their messages pointed to AI’s next phase: faster capability gains, bigger real-world rollout, and much higher stakes. AI is moving from impressive demos to something that sits inside infrastructure, jobs, and daily life. That is why the focus kept returning to the basics: access to compute, the data centre backbone, clean and reliable power, and real tests for safety, livelihoods, and language. In other words, the evolution of AI will be shaped less by who has the flashiest model, and more by who can build it responsibly, deploy it widely, and make it work for everyone.