Topic-Wise Conversation: Sam Altman x Nikhil Kamath

Sam:
The new system is an integrated model, which makes it more powerful and effective than previous versions.

It’s like having PhD-level experts in every field available to you 24/7—not just to answer questions, but also to perform tasks on your behalf.

For example:

Nikhil’s Question:
If I were a 25-year-old living in Mumbai or Bangalore, what should I do now?
What should I study or focus on learning?
If I were to start a company, what kind of business makes sense?
If I were to get a job, which industries will have strong growth over the next 3–5 years (not just 10 years into the future)?

Sam’s Answer:
This is probably the most exciting time in history to start a career.
A 25-year-old today has opportunities no previous generation had—whether in starting a company, becoming a programmer, joining an industry, or creating new media.

With today’s tools, a single person can achieve what once required decades of experience or large teams.

Industries with big potential:

Nikhil’s Question:
Is there anything specific you’d like to share about GPT-5?

Sam’s Answer:
GPT-5 offers a new level of capability, robustness, and reliability.
It can be used for many real-life tasks—creating software, answering questions, learning new skills, and working more efficiently.

India is now our second-largest market, and it may soon become the largest.
The team has incorporated extensive feedback from Indian users to shape the model’s improvements.

Nikhil’s Question:
As a 25-year-old in India, what are some low-hanging fruits I could build on top of GPT-5 today—areas or opportunities I should definitely look at?

Sam’s Answer:
You could build an entire startup far more efficiently than ever before.
With GPT-5, even a single person—or a small team of friends—can handle tasks that once required large teams and diverse expertise.

GPT-5 can help you with:

Essentially, it allows a 25-year-old today to do things that would have been impossible without big resources in the past.

Nikhil’s Question:
I understand the potential of AI in science, but what should I study to take part in this?
For example, if I’ve been studying engineering, commerce, or arts—
Is there a specific field of study I should focus on to use AI in developing something for science?

Sam’s Answer (Key Points):
The most important skill: get really good at using new AI tools (become “AI-native”).
Learning itself is valuable—develop the meta-skill of learning to learn quickly.

Whatever field you’re in (engineering, biology, arts, commerce), being fluent with AI tools will give you a big advantage.
In the past, the high-leverage skill was learning to program; today it’s learning AI.

Other key general skills:

Specific subjects like biology or physics are less critical right now compared to mastering AI fluency and general learning skills.

Nikhil’s Question:
When you say we should adapt, change, and learn AI tools faster, is there a specific path or approach to do that?

Sam’s Answer:
The best way to improve is by actively using AI tools.
GPT-5 is especially good at helping create small pieces of software quickly—better than any model before it.

Even in his own experience, Sam has been surprised by how often he’s used GPT-5 recently to build quick software solutions for everyday problems.

Nikhil’s Comment:
If the world of tomorrow were an “AI kingdom,” you’d definitely be some kind of prince.

Sam’s Response:
When I was 19, I thought leaders of big tech companies had everything figured out—that there were “adults in the room” with a solid plan.
Now that I’m supposed to be the adult in the room, I realize: no one truly has a perfect plan. Everyone is learning as they go.

If someone believes “I know better and the world doesn’t,” they usually make worse decisions.
What really matters is having an open mindset, curiosity, and the willingness to adapt to new data and change your mind.

The best founders Sam has observed in his career are the ones who are quick learners and highly adaptable, not the ones who think they already know everything.

Nikhil’s Question:
Do you think the world will keep capitalism and democracy as it is? For example, what if one company—say OpenAI—ever controlled 50% of global GDP? Would society allow that?

Sam’s Answer:
I don’t think that will happen.
AI will be much more distributed, not concentrated in one company.

Best analogy: the transistor—a breakthrough that seemed like it might capture all value but instead became embedded everywhere, powering countless products and industries.

Similarly, AI will spread across companies, products, and services, creating broad value instead of a single monopoly.
I once worried one company could dominate, but now I see that as naïve thinking.

Nikhil’s Question:
If something like AI gets very large, do the chances of the world moving towards socialism increase? Would it be nationalised?

Sam’s Answer:
I’m not sure about nationalisation, but I do expect more redistribution and social support as society grows richer and technology advances.

We’ll likely see experiments with sovereign wealth funds, universal basic income, and redistribution of AI resources.

Nikhil’s Question:
Worldcoin was an interesting experiment around universal basic income. What’s happening with it now?

Sam’s Answer:
The idea is to recognize humans as unique and special in the age of AI.
Worldcoin aims to create a privacy-preserving way to verify unique humans, then build a new network and currency around that.

It’s still early, but the project is growing quickly.

Nikhil’s Question:
We had discussed sectors like an “older and sicker world” and “gateway luxury brands” as discretionary spending rises. But what happens to these brands in a deflationary world, where purchase values decline?

Sam’s Answer:
In a deflationary world, some sectors face strong pressure, but others become sinks for excess capital.
Luxury brands might not decline—in fact, they could go up, as surplus capital needs somewhere to flow.

Nikhil’s Question:
Do wrappers built on large AI models—like Harvey—survive long term, or do they get replaced by the models themselves?

Sam’s Answer:
Some wrappers will survive, others won’t—it depends on the company’s choices and execution.
Simply using AI doesn’t create a defensible business.

Every tech boom shows the same mistake: people assume new technology removes normal business rules—but it never does.
You can build something amazing with AI, but you still need to create a real, defensible layer (brand, product moat, ecosystem, etc.) around it.10. Worldcoin

Nikhil’s Question:
Worldcoin was an interesting experiment around universal basic income. What’s happening with it now?

Sam’s Answer:
The idea is to recognize humans as unique and special in the age of AI.
Worldcoin aims to create a privacy-preserving way to verify unique humans, then build a new network and currency around that.

It’s still early, but the project is growing quickly.

Nikhil’s Question:
If I build a business on top of your model, should I worry that—like Amazon creating white-label brands—you’ll eventually cannibalize my business as you expand into many areas?

Sam’s Answer:
Think of AI like the transistor—a general-purpose technology that keeps improving and integrates into many products.

If your business gets stronger as the model improves, you’ll keep doing well.
If your business weakens as the model improves (because it’s just a thin wrapper), that’s risky—similar to past tech revolutions.

Many companies are already creating real value and strong customer relationships on top of AI.
Example: Cursor, a startup building with AI, is rapidly growing while forming durable bonds with users.

Nikhil’s Question:
Is the relationship stronger when I sell a service on top of your model versus a product—since services involve repeated interactions and room to add “taste”?

Sam’s Answer:
Yes, generally services allow for deeper relationships than one-time product sales.10. Worldcoin

Nikhil’s Question:
Worldcoin was an interesting experiment around universal basic income. What’s happening with it now?

Sam’s Answer:
The idea is to recognize humans as unique and special in the age of AI.
Worldcoin aims to create a privacy-preserving way to verify unique humans, then build a new network and currency around that.

It’s still early, but the project is growing quickly.

Nikhil’s Question:
If models become predictive of my style or behavior, wouldn’t my value go down unless I act in a contrarian way (different from my own past patterns)? Does the future favor contrarian behavior?

Sam’s Answer:
Good point. Yes, being contrarian and right will matter more.
Most contrarian ideas are wrong, but the rare ones that are right will be especially valuable.

The ability to do what models cannot easily replicate will only increase in value over time.

Nikhil’s Question:
Besides being contrarian, is there anything else I could do that AI models will take longer to replicate?

Sam’s Answer:
Models will eventually be smarter than us, but human connection matters deeply.

Even if AI could host a technically better podcast, it likely wouldn’t be more popular than a human host.

People care about life stories, personality, cultural identity, and shared social experiences.
We are fundamentally obsessed with other humans—that gives unique value which models can’t fully replace.

Nikhil’s Question:
Why do you think humans will still matter more, even if AI gets smarter?

Sam’s Answer:
It’s deep in our biology to care about real people.

In a world flooded with unlimited AI content, the value of being a genuine human will only increase.

Nikhil’s Question:
What’s the difference between AGI and human intelligence today and in the future?

Sam’s Answer:
GPT-5 today: superhuman in knowledge, recall, and pattern recognition for tasks lasting seconds to a few minutes.

Limitations: Not close to humans in asking the right questions or sustaining effort over long periods.

Progress example (math):

But proving a new important theorem might take ~1000 hours—we’re not there yet.

Nikhil’s Question :
In AI, the US seems a few years ahead. In robotics, China seems ahead. What’s your view on robotics?

Sam’s Answer:
Robotics will be incredibly important in a few years.
One of the most AGI-like experiences will be seeing robots walking around doing normal, day-to-day tasks.

Nikhil’s Question 2:
Do robots need to have a human-like form?

Sam’s Answer:
Not strictly necessary, but…
The world is built for humans (door handles, cars, factories, tools, etc.).
Therefore, humanoid robots make sense for most tasks.

There will also be specialized robots with non-human forms.

Nikhil:
As a young Indian founder, how do I compete with companies that already have manufacturing scale?

Sam:
In the long run, once robots are built, they could self-replicate (robots making robots).

In the short term, you’ll need strong manufacturing partners.
Even OpenAI is learning this — it’s a new skill set for them too.

Nikhil:
What happens to form factor (phones, computers) in the AI age?

Sam:
Traditional devices (phone/computer) are binary: either on (in hand) or off (in pocket).

AI works best when it has continuous context and can act proactively, not just when you “turn it on.”
That requires new form factors beyond phones/laptops.

Think: ambiently aware hardware → always present, more like a companion.

Nikhil:
Is the Jony Ive collaboration about a sensor/device?

Sam:
They’ll try multiple products.
Goal: build hardware that an AI companion can embody itself in.

This “AI companion form factor” will be an important thread in the future.

Nikhil:
You’ve invested in Helion (fusion) — does that solve climate change?

Sam:
Fusion would be a big step forward, but…
Climate damage is already done → we’ll still need ways to undo harm (not just stop emissions).

So, fusion isn’t the complete solution, but it’s a major enabler.

Nikhil:
What’s in this AI revolution for India as a country?

Sam:
India may soon be OpenAI’s largest market.
The enthusiasm + adoption of AI in India is unmatched globally.

AI lets India leapfrog into the future — skipping legacy stages of development.
India is one of the most excited and energetic societies in transforming with AI right now.

Sam: “I’m looking forward to coming soon. The momentum in India is unmatched.”

Nikhil:
How do we transition from being a consumer of AI to being a producer (building products the world uses)?

Sam:
That shift is already happening.
India’s entrepreneurial energy in AI is visible and growing fast.
OpenAI (and Sam) see India as a hub for AI builders, not just users.