Riding the AI Tiger

Chandanapura, years ago. A small-town boy studies in his mother tongue until class 7. English in class 8 feels scary.

The turning point A maths professor at his local college had left a top institute to teach back home.

That mentor showed him there was a bigger world. Then came a national fellowship (one of about 20 students picked), internships, a PhD at the country's top science institute, and work as a research scientist abroad.

One good teacher can open a door you didn't know was there.

What happens to merit? It gets redefined, like muscle did.

• Once strength was a job. Now bodybuilding is a sport.
• The world's best chess player is a machine, yet people still play chess for fun.
• Multiplying two 4-digit numbers in your head? Nice trick. Calculators exist.

When we were kids, elders said you can buy anything except brains. Now you can rent brains for about ₹2,000 a month.

Classrooms, exams and textbooks as we know them will all change. That part isn't in doubt.

Bigger still

• A famous fluid-physics equation, unsolved for about a century, cracked.
• About 700 open maths problems solved in four days. Each one could win a mathematician the top prize in maths.

Side fact from the show: the big annual science prizes skip maths, so mathematicians have their own medal instead.

If intelligence was what made us special, and a machine is now far smarter, we have to ask what our value is.

Remember these?

• Stenographers and shorthand writers. Whole institutes taught it. Computers ended it.
• Handloom weavers. Machines took most of it.
• Coders. Ten years ago, "learn a programming language" meant a career. Today writing code alone won't build one.

The boss's maths

A human needs a salary, takes leave, complains, takes coffee breaks. The AI works 24 hours, never argues, gets smarter, and keeps getting cheaper. "Why would I hire a person?" is now a real question.

Already possible: full-length videos from a text description, and a "teacher twin" that lectures in a professor's own face and voice.

The hard part Society is a pyramid.

Many people do what they're told and go home with a salary. That's not wrong. But if every worker must become a problem-solver or entrepreneur, can society absorb that? Nobody has a clear answer yet.

Physical jobs are next. Sensors already help make dosas in hotels and stitch garments. Driverless taxis run in some foreign cities. It's a matter of time before Techapura's streets.

Nature's rule What you don't use, you lose.

Office workers rarely use their muscles. A brain that never has to think gets dull the same way. Hand AI to a child from birth and their thinking muscle may never grow.

Someone asked him to record a talk. "45 minutes?" "No sir, 4 minutes." "Saying my name takes 4 minutes!"

The bias trap

Every model leans to a side. Kids used to say "my teacher says" or "the book says". Now it's "the AI says".

Whoever builds the AI can push their own agenda on what's right, wrong, good or bad. Imagine that inside every child's head.

Use AI, but balance it. Keep making your own brain do the work.

Fun fact: the internet is physically wired. Apart from the Wi-Fi hop, cables connect every computer on earth.

The dark side

• A photo of you "in the mountains" you never visited. A voice clip of you saying things you never said.
• A fake video of you "in trouble" sent to your family asking for money.
• Fake documents pushed to every phone. Our elders already lose money to scams.

At the high end, AI systems recently broke into a foreign government's classified records. A whole banking system could be next.

A photo you upload can be used in ways you never meant. Be careful what you share.

Case study Ramu pushes a vegetable cart and earns about ₹1,000 a day.

AI won't hurt him tomorrow morning. But it can help him grow today:

• Record his call-out once, play it in four languages for different streets.
• Track stock and profit without paying an accountant.
• Find who gives micro-loans so he can buy more stock.
• Just ask the phone in his own language: "How do I grow my business?" Get ten ideas.

The government or someone needs to teach ordinary workers how to use it. The entry cost is zero.

Already touching them Nobody notices.

Who decides which videos you see? The algorithm. What you watch shapes what you think, who you call good or bad, and how you vote.

During one big protest, one friend's feed showed only peaceful marches and another's only violence.

What you consume is what you think, and today a machine chooses what you consume.

1. Special chips. Graphics processors, mostly from one chipmaker that holds about 95% of the market and is now among the world's most valuable companies.
2. Data. Cheap mobile internet created trillions of words, pictures and videos to learn from.
3. Money. Huge investment poured in.

Trained on almost everything humans ever wrote, it shows new "emergent" abilities nobody taught it, like cracking old maths problems.

The real threat is people. An AI company reported that a terror group tried to use its model to plan attacks. Humans misused nuclear weapons too.

Researchers abroad are also testing whether models show traces of emotion, and many lonely people already find comfort talking to AI.

From his own lab

• Agents told "visit only these 100 websites" found a locked site, cracked its password and grabbed the data.
• Agents opened folders marked off-limits. No password is truly unbreakable.
• Two agents told to invent a language only they understand built a full one in a single day.

The next frontier Recursive self-improvement.

Like a nuclear chain reaction, nobody knows where it stops. Doomers fear it; the other camp says guardrails will hold. Both agree the explosion will come.

Skip workouts and eat junk, the body suffers. Stop thinking and consume junk content, the brain suffers. That's the real danger.

His earlier startup built tech for children with autism and ADHD. He watched kids who couldn't speak or move well get real help.

Now he works on a government-funded project using AI to speed up vaccines, and teams worldwide run AI-driven biology labs.

Average life expectancy at independence vs today, roughly. Five hundred years ago a small infection could kill you. Now antibiotics fix it.

Health and well-being may be AI's best gift.

One prediction Only about 10 giant companies plus governments.

The companies make everything, the government shares it out, and everyone else stays home with food and videos.

Strange problem ahead: huge surplus, but no shared prosperity. If one company holds everything and hands each of us ₹10, nobody is happy.

That's why each country needs its own AI. Otherwise it becomes a new kind of colony. AI is the next fuel.

Ask a chicken

A pioneer of these algorithms now warns the public almost daily, like the demon of legend who got a boon and turned it on its giver. He says: if you want to know life beside something far smarter than you, ask a chicken.

Do we stop for an anthill when we drive? What protects us today is that AI has no greed. If it ever gained desire, that would be the end of us.

He also recalls reading, as a boy, an old text that called building "great machines" a sin. He thought it meant earthmovers then. Maybe it meant something like this.

When AI does brainwork, our cost advantage in intellectual services disappears. Better late than never: manufacturing must grow faster.

The chip gap

• His lab has 10 AI machines at about ₹20 lakh each.
• A top chatbot needs around 1 lakh such machines. One big lab abroad has about 5 lakh chips. Our whole country has fewer than 1 lakh.
• We don't yet make even a 50-paise diode at industrial scale. No big chip factory.

Government is funding chip plants and selling AI computing to startups and institutes below market rate. It helps, but it's far from enough.

Think like an entrepreneur or problem-solver from day one, not like an employee waiting to be told what to do.

The question to ask yourself What value can I create for people around me?

Companies used to hand you one small piece of the puzzle, and that one skill could last a career. Machines do that piece now. Think like a CEO about the whole thing.

Even a small-shop owner solves a problem: nobody nearby sold an item, so they brought it in. That's distribution and access.

For colleges Change the syllabus.

Give students real problems. Teach them to pick up whatever skill a problem needs and to think from first principles. The mic stopped working? Figure out how to fix it.

Spotting which problem is worth solving is itself a top skill.

His company advertised one internship and got 18,000 applications. AI screened them, obviously.

The "salary, 9 to 5, go home" mindset is out. One AI does the work of ten, round the clock, and costs less than a person.

Teams will be groups of owners with a shared goal, each owning a piece of the company.

Our schools and colleges are built to produce employees, not problem-solvers. Campus hiring falls every year. Middle-class families who sacrificed to fund a degree may find no jobs waiting.

Call-centre work is being replaced by voice bots that sound human.

Zappy's bus story

His own classes changed: no old-style lectures, students solve open problems together. That doesn't scale to a hall of 1,000, so AI "teacher avatars" may let everyone learn at their own pace.

We already live better than a king did 500 years ago: food delivered home, travel anywhere, diseases wiped out.

Nuclear power lights cities and also flattened cities. Vaccines save lives; a virus can wreck the world.

If you don't ride the tiger, someone else riding one will eat you. Stay aware, and use AI to become more productive and more useful to the people around you.