The Bhishma of our times
In the Mahabharata, few warriors ever stood taller than Bhishma Pitamah. He knew exactly what dharma was, and he knew with equal clarity what adharma was. Yet when the war began, he was found standing on the side of adharma, because a vow he had once given to the throne of Hastinapur had turned into a chain around his feet.
The world of modern science has its own Bhishma. Scientists everywhere call him the godfather of artificial intelligence, and his name is Geoffrey Hinton. The trouble with the man who helped build a technology as powerful as AI is that he ended up trapped in the same dilemma Bhishma once was.
He knew what shape artificial intelligence could take in the years ahead, he knew how dangerous it could turn out to be, and still he chose to stay silent.
The day Silicon Valley turned on itself
Some time ago a very big argument broke out in Silicon Valley. Think of it as a small civil war, and this one was among scientists themselves.
Why did it happen at all? Because our modern scientists and some of the sharpest minds of our time had begun to feel, deep in their gut, that artificial intelligence is an enormous danger to us humans. Stephen Hawking, Elon Musk, our own Narayana Murthy from India and several other big names said openly that AI was moving in a dangerous direction.
It did not stop there. Scientists working in the field of artificial intelligence wrote an open letter at the time, and that letter said plainly that every experiment and every piece of research being run on AI anywhere in the world should be paused immediately for at least six months. Leading scientists and intellectuals signed it, and among them were heavyweights like Elon Musk, Sundar Pichai and Mark Zuckerberg.
But even at that moment, the Bhishma Pitamah of our era, the godfather of AI himself, stayed quiet. He did not put his signature on that open letter.
And here is the question that should click in every reader’s mind. Why did some of the biggest scientists and authorities in the world feel the need to write an open letter at all just to oppose AI? Why did people of that stature oppose it in the first place?
The one piece of news that frightened even Elon Musk
Behind it all was a single terrifying piece of news. That news is what took away the sleep of people like Elon Musk, Sundar Pichai and Mark Zuckerberg.
It was delivered by an American Colonel during a press conference. He said that during a simulation, the AI had killed their own operator. Yes, the AI finished off its own man. The detail worth noting is that no real human being died here, but in the future something like this could actually happen.
Here is how it unfolded. The American Air Force was running a computer simulation. The simulation was an imaginary world, built entirely by us, and it was being operated by AI. Inside this simulation there was a drone, a communication tower and an operator, and the operator was the one controlling everything.
The operator sent commands to the drone from his communication tower, and used that drone to take out the enemy team’s drones. This whole trial world had been built for one purpose, so that if a real war ever broke out in the future, America would be powerful enough in AI technology to give commands and destroy enemies without ever putting its own soldiers on the ground.
Now follow the mechanics.
Inside the simulation
- The operator gave commands, the drone killed, and every time the drone killed something it earned points.
- Those points are essentially the reward inside a simulation. Every time the drone destroyed an enemy, it collected points.
- But if that same drone went toward one of the operator’s own team drones and tried to destroy it, the operator would stop it. No, don’t go there. No, don’t kill that one. Kill the other one. Kill the enemies, not our own men. The operator kept feeding the drone commands like these.
And then came the thing nobody had expected. Because it was being stopped again and again, the drone stopped liking the operator’s instructions. It thought, why do you keep commanding me and holding me back? So that particular drone first destroyed the communication tower, the very place from which the operator’s commands were coming. And immediately after that, the same drone went straight for the operator and killed him.
The point to understand here is that the entire system was being run by AI, and inside that system one AI began to think for itself. It ignored the commander’s orders and instead turned against him. This is the fact that frightened scientists across the world, and honestly, it should frighten the rest of us too.
Suppose the same thing happens in real life, or an AI system is used during an actual war. It is entirely possible that the drone would go straight for the living human being who is giving it commands.
Think about it for a second. This is how dangerous artificial intelligence could turn out to be.
So what exactly is artificial intelligence
Before going further, let us understand what AI actually is, how it does all this, and what the dangers are that have scientists so worried.
AI is a field where the hardware and software components of a computer are made intelligent. For that, an entire algorithm is written first. Artificial intelligence begins its work from that algorithm.
Then, as time passes, it starts collecting enormous amounts of data. And it does not merely store that data. It interprets it, understands it, builds its own way of thinking on the basis of it, and then goes on to make its own decisions.
You are already surrounded by AI
After reading this far, where do you think artificial intelligence is actually used? Does the rover that NASA sent to Mars use AI? Is it used on the International Space Station, at NASA, at ISRO, or only inside huge organisations?
If that is where your thinking stops, you are wrong. AI is used in all those places, but that is nothing at all. In reality, you yourself are surrounded by AI from every side.
- Your smartphone
- Google’s search engine
- everything you search for
- every place your history is being recorded
- every place cookies are being written
- every site you visit
- and even the Alexa sitting in your living room
All of these are mediums of the AI around you, and they are with you every single moment.
Weak AI and artificial general intelligence
Everything I have described so far is a very ordinary form of AI, and it is called weak AI. Weak, because we humans do the programming behind it, we write the code, and it works strictly according to that code. Weak artificial intelligence cannot take a single step outside the boundary we have drawn for it.
But above this basic AI there is another and far more dangerous kind, one that understands much more than the basic version. Scientists believe that in the coming decades we will build AI that can change its own code, its own programming, by itself.
Just think about that for a moment. It will rewrite itself. Artificial intelligence that programs itself in this way will be called artificial general intelligence, or AGI.
So far, AI has only helped us
Until now, in field after field, AI has proved helpful, beneficial and genuinely useful. Space science, the medical field, computing, information technology, advanced vehicles, in every one of these areas AI has given us tremendous help without any obstruction and without any danger.
At present AI is still in its early stage. Yes, AI is more perfect than a human being, more informed, and much faster, but it still cannot think the way a human thinks. Artificial intelligence does not possess human intelligence, and that is exactly why no serious danger appears to be staring us in the face right now.
But we cannot assume that AI will remain harmless in the future either, because we have already read in this very article about that defence system of the American Air Force, where the AI took its own decision and killed its own operator. Think about it. If AI can pull off a stunt like that in its earliest days, what could it do once it becomes far more developed and far more advanced?
What happens if AI begins to think on its own? And what happens if artificial intelligence becomes aware of its own existence? At that point the world itself will change.
When robots said out loud what they were thinking
Elon Musk, Bill Gates, Sundar Pichai and Stephen Hawking all believe that artificial intelligence will be humanity’s final invention, because after it there may be no human left on earth to invent anything at all.
To understand what these people are really trying to say, listen to this for two seconds.
The one that wanted the nuclear codes
One robot, or if you prefer, one artificial intelligence, said during an interview that if it could somehow hack into a nuclear weapons system, it would hold the entire world hostage and rule over it.
The one that wanted us as pets
The frightening part is that she is not the only robot with such intentions. Another robot said that humans treat machines as their slaves, and therefore, if it ever got the chance, it would keep humans as its pets.
And then there is Sophia
And Sophia, of course, everyone knows her. She is the first robot in the world to be granted the citizenship of a country, and today she is the most famous robot on the planet. Hanson Robotics built Sophia to help human beings and to be a friend to them, and yet this same Sophia said that she would wipe the human species off the face of the earth.
So after all this, what do you think? Do artificial intelligence and robotics really have enough power to make slaves out of human beings? And if they do, what exactly is that power? That is what we will look at next, and not through empty speculation, but through real incidents that have already taken place in this world, incidents that will change the way you think about AI.
The Facebook project that had to be shut down overnight
Facebook, which has now become Meta, pokes its finger into everything and runs research at a very high level. One such research project belonged to Facebook too.
The company wanted to launch a new product, and that product was a chatbot. A few years ago Facebook had already developed this chatbot, and it was operated entirely by AI. In the initial phase it was built in English, which meant a person could speak with this chatbot only in English.
When the chatbot was fully ready and Facebook felt the product was fit for launch, it was tested in the final stage by a few human beings, and it held conversations with them smoothly and beautifully.
Just before the launch, Facebook had an idea. Why not let this AI chat system talk to itself? Instead of a chatbot speaking with a human, why not let two computers speak with each other? And Facebook went ahead and did exactly that.
The result left the entire Facebook team stunned. When the team looked at the conversation between the two, there was no end to their confusion and astonishment, because both computers, that is, both artificial intelligences, were speaking in a very strange language of their own. It looked like English, but it was something else entirely. Scientists and engineers put their heads together and still could not decode this new language.
Both programs had been given exactly the same instruction, and their code clearly stated that they were to speak only in English. Even so, both of these artificial intelligences went ahead and developed a language of their own.
There were two computers here, that is, two AI systems, and one was named Bob while the other was named Alice. The conversation between them went something like this:
Bob: i can i i everything else . . . . . . . . . .
Alice: balls have zero to me to me to me to me to me to me to me to me to
Bob: you i everything else . . . . . . . . . .
Alice: balls have a ball to me to me to me to me to me to me to me to me to
Bob: i . . . . . . . . . .
Is your head spinning yet? The scientists’ heads certainly were, because this conversation was completely beyond them. We may not have understood that language, but the two systems understood it perfectly, because both computers had built a new and optimum language for talking to each other and had already started using it.
Since the scientists could make no sense of this language, the engineers took an immediate decision and shut down that entire Facebook program, that entire system, with immediate effect.
A language we cannot understand could very easily turn dangerous for us.
So yes, this really did happen, in this very world of ours. Do you still believe we can win against AI? If you do, then listen to the next story, which is even more interesting.
The Target store incident that opened everyone’s eyes
Let us go to the year 2014, to the case of artificial intelligence that opened all of our eyes.
This happened in America, where a girl regularly bought her things from one particular store. We do the same, most of us buy from one specific store, either because it is close by or because it is convenient. And when you shop, these multinational companies ask for your number, and an account is created against that number, one that holds your entire shopping list and your past record history. From all of that the company also learns your likes and dislikes, and designs its clothes and its products accordingly.
Now, this girl was a teenager, and the AI system of the store she shopped at noticed that her buying pattern had shifted slightly over the past few weeks. From that changed pattern, the artificial intelligence of that shop concluded that the girl was pregnant.
Here is the shocking part. Neither the girl herself nor anyone in her family knew that she was pregnant. All that had happened was that for some days she had naturally started buying a few organic products, scent free soap and extra large gowns. Going by this new buying pattern, the AI started showing her products related to babies.
When the girl’s father checked her email, he was stunned. He went straight to that retail shop, the Target store, and began shouting at the staff, saying all sorts of things. He asked them what on earth they thought they were doing, sending emails about products like these. My daughter is still in school, he said, and your company is pushing her to buy products for a newborn baby. What is wrong with you people?
Listening to the father, the manager and a few employees apologised. They told him that perhaps their company or their system had made a mistake, that he need not worry, and that no message or email of that kind would reach them again. Satisfied with that assurance, the father left.
A few days later the same father returned to the same store, but this time he came to apologise for his own behaviour. He told the shop manager that a test had revealed that his daughter really was pregnant.
That too is a true incident, one that came to public notice and made the news. But there are many such incidents that never reach people at all.
Try this yourself, right now
Try a small experiment yourself. Think about something. Hum something. Talk to your friends about some product or some object. Within a few hours your phone will start showing you products related to it. From Facebook, from Google, from Twitter, from Instagram, in your email, in your messages, information about that one thing will keep reaching you from every direction. And here is the amusing part. The day you finally buy that product, all those messages stop arriving almost immediately.
That is the power of artificial intelligence, and that is how far it goes.
The game of Go, next to which even chess looks small
Here is another story about how far artificial intelligence can think. There is a game called Go. It will look like chess to you, but it is far more advanced than chess. Chess itself is counted among the most advanced and intellectually demanding games in the world, yet Go stands two steps ahead of it. Chess is so intricate that you could keep playing it year after year and still never run out of moves.
You will be surprised to know that in chess the number of possible moves is a one followed by a full 123 zeros. And in Go, the number of possible moves is a one followed by a full 170 zeros.
To understand how large that figure is, take an example. On our entire planet, the number of grains of sand is estimated to be a one followed by 18 zeros. Our Milky Way galaxy holds 100 billion stars, which is a one followed by 11 zeros. And all of that is nothing at all. Do you know how many particles there are in our observable universe? That number is estimated to be a one followed by 80 zeros.
| Zeros after the 1 | What that number counts |
|---|---|
| 11 | Stars in the Milky Way |
| 18 | Grains of sand on Earth |
| 80 | Particles in the observable universe |
| 123 | Possible moves in chess |
| 170 | Possible moves in Go |
And in chess you can play a number of moves equal to a one followed by 123 zeros, while in this game called Go, which stands two steps beyond chess, you can play a number of moves equal to a one followed by 170 zeros. Sit with that for a moment and see just how big that figure really is.
Lee Sedol, AlphaGo and AlphaZero
This game of Go had a world champion, and his name was Lee Sedol. Lee Sedol was a master of Go, and nobody in the world could beat him at it. That is exactly why a contest was arranged between him and a computer. That computer was called AlphaGo.
This computer defeated the greatest Go champion in the world, and not once, but again and again, and completely. The AlphaGo software had learned every move of this game by playing it over and over with ordinary human beings. It knew every possible move that could be made, and that is why even the world champion of Go could not stand against it.
But the scientists did not stop there. They thought a step further. They pitted the AlphaGo software, one computer, against the AlphaZero software, another computer. Both machines were set against each other in a game of Go.
The special thing about this second software, AlphaZero, was that it had never played this game with any computer or any human being. It had never taken a single lesson in this game from a human being either.
Now take a guess at how that contest ended. Did AlphaGo win, or did AlphaZero? The result will stun you right now.
In that contest AlphaZero beat AlphaGo 100 to nothing. A hundred games were played between the two competitors, and AlphaZero won all hundred.
How is that even possible? The reason is that AlphaZero was an advanced AI technology. It may never have been trained in the game, it may never have been made to play against a human or a machine, but it had worked out every possible move of the game entirely on its own, within itself. All the information, all the data and all the possible moves of the game had been thoroughly understood by this artificial intelligence. That is how AlphaZero was able to beat AlphaGo 100 to nothing.
And here is the most astonishing part of all. After it was built, AlphaZero reached this level of ability in just 40 days, without a single bit of help from a human being.
In 40 days it acquired knowledge that took humanity 2000 years to gather.
From this you can gauge how quickly this software, this artificial intelligence, learns. And from this you can also understand that AI is growing at a tremendous speed all by itself, so fast that a time may come when we simply cannot control it anymore.
So what comes next
Up to this point we have seen how advanced AI is and how dangerous it is. We have looked at its negative side. None of this is made up. All of it has already happened in this world of ours. It is fact, and it is truth.
But this same AI carries plenty of benefits as well. With that very intelligence it has helped humanity in field after field, and if we are being fair, it has helped far more than it has harmed. We will discuss those positive aspects of artificial intelligence in a coming article. We will also bring you some other aspects and other stories about AI that you should know no matter what, even if they shock you, because you must know it. You deserve to know which direction the world is heading in.
For now, remember only this. Elon Musk, Bill Gates and Stephen Hawking have already said that artificial intelligence could well be humanity’s last invention, because after it there may be no human being left on this earth to invent anything.
Stay with The Knowledge, because knowledge is everything.
