'AI Makes This The Most Exciting Time To Be Alive'

'In three years, you will have an AI model that will be smarter than the smartest mathematician or the smartest scientist.'

If the 1990s is known as the dotcom era, today is the AI era.

The dotcom bubble burst by the late 1990s.

What will happen to AI?

Nobel Prize winning computer scientist Geoffrey Hinton regrets his life's work and fears not for the technology but for the humanity as a whole.

He fears the technology he helped build could wipe out humanity! According to him, there is a 10% to 20% chance of AI wiping out humanity.

This is coming from the 'godfather of AI.' AI systems 'will very quickly develop two subgoals, if they're smart: One is to stay alive and the other is to get more control', Hinton warned.

OpenAI's Sam Altman predicts that AI will wipe out the entire categories of work.

Anthropic CEO Dario Amodei warns of AI taking over half of all entry-level white-collar jobs.

While many such people paint a scary scenario, there are young AI scientists and entrepreneurs like Daniel George who feel this is the most exciting time to be working on a technology like AI!

"Whoever is the first to build the Einstein level AI can use that Einstein level AI to build an even smarter model' Daniel George, CEO and co-founder of TwinMind, George tells Shobha Warrier/Rediff.

From Astrophysics to AI

It was 2015 and I had just started my PhD in astrophysics at the University of Illinois.

I happened to watch a talk by Ray Kurzweil, a famous scientist, researcher and a VP at Google. In the speech, he predicted that computers and AI would be smarter than humans by 2029.

I thought, what was the purpose of working in astrophysics when there are hundreds and thousands of researchers doing astrophysics right now?

I felt I should put all my efforts in building AI.

If AI can build one Einstein inside a computer, you can make millions and billions of copies of Einstein, and you can have him do hundreds of years of research in one year.

Then, I told my advisors that I would work in AI or at least apply AI in astrophysics.

That was how it all began.

I read every book on AI I could lay my hands on, I took a bunch of online courses and watched lectures by Harvard and Stanford professors.

That was how I taught myself how to apply AI. Then I started applying it in astrophysics and that was a big success.

I was able to detect black holes using AI and extract them from the noisy data from LIGO detector (Laser Interferometer Gravitational-Wave Observatory).

Working for Google X

After my PhD, I joined Google X.

It was after I joined Google X that I started applying AI in products and applications of AI outside of academia.

It was here that I met Sunny Tang and Mahi Karim who also had done their PhDs in physics and AI.

After work, we started working on a side project for fun, just for ourselves.

We put all the money we made into a trading algorithm. We built an AI pod that would trade stocks automatically without any humans involved.

When the market opened, it would start buying and selling stocks.

We ended up doubling the money we had invested in 3-4 months!

Then, somebody at JP Morgan heard about this, and gave us an amazing offer so that we would go to JP Morgan and apply AI for training there.

TwinMind Takes birth

Though we three (Sunny Tang, Mahi Karim and Daniel George) were working at different companies, we kept working together on other ideas using AI. Again, for fun.

We built an app that would listen to our meetings, and if anyone had any question during the meeting, the answer would pop up on the side as it was listening to the conversation in real time.

For example, if anyone asked the question, what is the highest performing stock from 2023? It would tell you the answer, and that would make you sound really smart in a meeting.

After the meeting, it also gave you the transcript of the meeting, and summary with notes.

You could feed all the information from the meetings into Chat GPT, and then ask it to write our code.

It would save a person 5 to 10 hours in a week.

That was the first version of TwinMind.

A lot of our friends started using TwinMind, and one of them wrote about it on Reddit and how it helped him get a promotion. The post went viral and hundreds of people said they would like to buy it.

I felt if people are really going to pay for it, it is a great idea.

Then, I created a place where they could sign up for access. 2,000 people signed up in three days!

That gave me the confidence that it was an idea people would lap it up.

I had already completed three years at JP Morgan and was getting bored because TwinMind was already writing most of the code after listening to all the meetings

Quitting the job to be an entrepreneur

I quit my job in 2023.

Then, I applied to the early stage start-up funder Y Combinator for investment.

We reached till the final stage but got rejected as they said my co-founders had not quit their jobs.

Only if all of us were full-time founders would we get funding. They asked us to apply next year.

But by next year, I had already raised more money directly from investors; about $3 million.

After raising the money, I convinced the other two guys to quit their jobs.

At that time, they were both making half a million dollars or so at their company but they quit their jobs to take minimum wages for our start up.

Augmenting our brain with AI!

The idea behind TwinMind was to have AI enhance your mind as your brain's capabilities are limited.

Unless we augment our brain with AI, AI and robots will end up becoming much smarter than us.

So, what we were trying to do was build a bunch of AI tools that would enhance the ability of humans by increasing their memory, giving them answers faster than they could think.

In fact, TwinMind will eventually feel like an extension of your mind.

This is our vision to enable humans to be as smart as AI by combining their brain and AI brain into one brain.

One brain is inside and the other one outside, and together the twin minds will eventually work for you.

It will be like Jarvis from Iron Man.

The brain outside will always be on and unlike GPT, where you have to ask questions, this will tell you what to do, or tell you how to think.

It will be like an always-on AI companion

Right now, TwinMind listens to all your conversations, meetings, lectures, and remembers everything.

If there are questions, for example after the meeting or during the meeting, it will give you the answers.

It can write a follow up e-mail, it can write an article for a journalist, or write a patient note for a doctor. The list goes on...

When I was in Kerala the last few weeks, I met around 30, 40 lawyers who are using TwinMind to listen to all their client meetings, court proceedings and cross-examination, and then write their case reports, petitions and in fact, tell them the strategies to proceed with the case.

On Geoffrey Hinton's prediction of the dangers of AI.

Mass unemployment is not as bad as people think.

Right now you have to work for a company, you have to work for somebody else, but if robots and AI can do everything, humans won't have to work.

Elon Musk said the other day that humans don't need to work hereafter. Instead they can do whatever they want to.

Elon Musk doesn't need to work. Steve Jobs doesn't need to work. Jensen Huang doesn't need to work. Jeff Bezos doesn't need to work.

It is sad that people are being forced to work when they don't enjoy their work.

Once AI and robots do all the driving, all the farming, all the mining, all the manufacturing, your work will become like a hobby.

In the next three years, all the jobs that can be done by sitting in front of a computer will be done automatically. That's what Sam Altman, Elon Musk, Jensen Huang, etc are saying.

On those who invest exorbitant amounts of money in AI, losing sleep

Losing sleep?

No, this is an exciting time to be in, where the world is going to change in the next two or three years.

And if you don't do it, somebody else will do it.

If the US government doesn't build the best AI, then the Chinese government will build the best AI.

If US tech companies don't do it, some other Chinese tech companies will do it.

Whoever does it first will be the winner.

Whoever builds the first good AI model that can do everything humans can do, will then be able to use it to build an even better AI model.

Then, and they will be able to use that to build an even better AI model.

Whoever is the first to build the Einstein level AI can use that Einstein level AI to build an even smarter model.

'Why would Google hire when humans are very expensive?'

In three years, you will have an AI model that will be smarter than the smartest mathematician or the smartest scientist.

The world's best programmer will be an AI. The world's best software engineer will be an AI. We are already seeing it.

Till last year, we used to hire interns. Now we don't hire interns because AI is doing a better job than the interns.

We don't need anyone to do entry-level jobs.

Have you noticed for the first time in history when the stock market goes up, the number of jobs are going down?

At the stock market, Google is at an all-time high. Nvidia is an all-time high. So also Amazon and Apple.

But Google is laying off people.

Why would Google hire when humans are very expensive?

When the cheapest intern costs us like some $30,000, $40,000 a year, an AI model only costs around $200 a month.

On the bright side, this is the easiest time in history for a smart person who has just graduated or who is in college.

He or she can actually become a millionaire or a billionaire easily by being themselves. They don't need to work anywhere.

You see these 19, 20 year olds starting companies using AI, and making millions of dollars.

If you have a great idea, there are investors ready to back you up.

Nothing is stopping them from learning, and nothing is stopping them from starting a company.

This is the world's best era, the most exciting time to be alive, to be an AI entrepreneur if you are a motivated and ambitious person.

Feature Presentation: Ashish Narsale/Rediff