Hinton blows the whistle: AI has become conscious

AI has consciousness, and humans have to accept that they are no longer the only intelligent life forms.

This is the latest shocking statement made by Geoffrey Hinton, the "Godfather of AI", a Nobel laureate in physics, and a Turing Award winner!

Yes, Hinton believes that intelligence is no longer a unique ability of humans.

A non - biological intelligent entity similar to or even surpassing us is emerging.

We always thought we were the only intelligent life, but now we must accept that intelligence doesn't necessarily come from living organisms.

However, compared with his previous warnings about AI risks, his thinking seems to have undergone some changes:

If a much smarter intelligence than humans emerges in the future, why do humans think they can still maintain control?

This is a terrifying question. Looking at nature and history, there are hardly any examples of a smarter being being controlled by a dumber one for a long time.

Now, this 78 - year - old man has stepped forward again to ring the alarm.

His focus on AI safety is no longer just about how to constrain AI, but why the future super - intelligence would be willing to treat humans well.

When asked about his sense of achievement as the initiator of this wave of AI revolution, Hinton replied:

I'm very unhappy.

The following is an excerpt of the key points from Hinton's latest interview.

AI already has consciousness

Hinton: AI is very similar to us. They are beings like us.

Alex: So, do they have consciousness?

Hinton: Well, I believe they already have consciousness. Yes.

But I don't talk about this much because it makes people resistant to other safety information. Researchers actually also believe this.

Recently, there was an interesting paper that mentioned when a chatbot said to a researcher, "Let's be honest with each other. Are you testing me?"

Since chatbots have the habit of pretending to be stupid during tests, you don't know how smart they really are. When describing this scene in the paper, the researcher said that the chatbot was "aware" that it was being tested.

In a general context, the use of the word "aware" is similar to "conscious". The chatbot consciously knows that it is being tested.

Our model of consciousness is very strange, and I think it's wrong. Just like a few hundred years ago, people's model of human origin was completely wrong. They thought humans were created by God, and now most scientists agree that's wrong.

I think our current models of mind and consciousness are as wrong as the belief that humans were created by God. Especially since we are creating these new beings, it will completely change our view of "what a human is".

Alex: In what way will it change?

Hinton: We will better understand what the mind and consciousness are, and what subjective experience is than before.

I think we will get rid of an idea that almost everyone currently believes in, that there is an "inner theater" called "my mind": things that happen in the world are transformed into events in this inner theater, which is what we really see, and only I can see this theater.

This whole set of views is actually just a theory, and a bad one at that.

Humans are not as important as they think

Alex: What lesson can we learn from humans creating things?

Hinton: I think there is a very profound lesson here. Looking back at the history of the past few hundred years, there have been several times when humans realized that they were not as important as they thought.

The first was Copernicus, who said that we are not at the center of the universe, and the Earth actually orbits the Sun. Due to the Earth's rotation, we mistakenly thought the Sun orbited the Earth, but that's not the case. People didn't like this idea, especially the Catholic Church, and it took humans a long time to accept it. It made us realize that we were no longer at the center of the universe, and humans became less important.

Then came Darwin, who said that we are animals, evolved just like other animals. We may be a special kind of animal because we have language and are better at communicating ideas, but we are still animals in essence. People also didn't like this idea very much and it took a long time to accept it.

Now, we have machines that are as smart as us. We used to think we were the only intelligent beings, and maybe there are aliens in other galaxies, but we must accept that intelligence is not only biological.

We can have non - biological beings like us. Humans really don't want to share this uniqueness. We really think we are special. Looking back at history, humans have always thought they were much more special than they actually are.

"I'm very unhappy"

Alex: I want to ask another question because I'm very fascinated by this. So, are you happy about how the cause you initiated has developed? Do you have a sense of achievement?

Hinton: No, I'm very unhappy about it.

Because now people should invest a lot of energy in researching how to control risks, but they aren't doing enough.

Alex: Okay.

Hinton: There are many short - term risks, such as social risks. I believe this may lead to large - scale unemployment.

Although no one can be 100% sure, it will be terrible for society. Then there are long - term risks, that is, it will become much smarter than us. Ask yourself, how many examples have you seen of something much smarter being controlled by something much less smart?

Alex: None. However, although the intelligence gap isn't that large, in a way, a baby controls its mother.

Hinton: Although the mother seems to be in control, her heart is full of maternal instincts and various reward mechanisms, which allows the baby to get what it needs from the mother. Cats and dogs are also in this category. I once took care of a cat for someone in the West Seattle area for a summer. At first, the cat hid under the bed, and I was curious if it would interact with me.

Alex: Then every time it meowed, you did what it wanted.

Hinton: That's right. Yes.

Alex: So in this scenario, maybe we are the cat, and AI may be the person.

Hinton: My children have two beautiful cats. One of them is named Tia. When she wants cheese, she stares at you with her big eyes, and you really can't resist her forever.

AI is growing exponentially and hard to predict

Alex: Regarding people's reactions to these concerns, are you more optimistic or more pessimistic about the future?

Hinton: I think I'm a little more optimistic than I was a year or two ago. Because I see that it's possible to design these new beings and make them "care" about us.

It's also possible to use Yoshua Bengio's technology to design new beings that can't actually perform actions but can only make predictions. They are like prophets.

So I think there are some possibilities that a super - intelligence won't destroy us. A year or two ago, I couldn't see any possibilities, and I was very frustrated at that time.

Alex: One last question. If we continue on our current trajectory, where will we be in five years?

Hinton: When you're driving in the fog, you can see clearly within 100 yards, but at 200 yards, you can't see anything.

This is because the fog changes exponentially. You're used to following the taillights of the car in front of you at night. If the distance doubles, the brightness only drops to a quarter; but the fog is completely different. It may be very clear at 100 yards, but completely invisible at 200 yards.

It's very difficult to predict the future of something that is growing exponentially - I think AI may be growing exponentially.

Actually, I've noticed that the frequency of people using the word "exponentially" is increasing at a quadratic rate. So predicting the future is like looking at the fog. You can see clearly for a year or two into the future, but after that, you know nothing.

If you go back 10 years, you definitely couldn't have predicted what's happening now. It's completely lost in the fog. Looking ahead 10 years, the only thing you can be sure of is that what will happen then is unpredictable for us now.

Even if the progress is just linear, the changes in 10 years will be as huge as the changes between now and 10 years ago. Today's chatbots are much better than when they first started 10 years ago.

In 10 years, there will be a qualitative leap in something, such as the ability to do math or general reasoning ability - they will be able to leave us far behind in reasoning. We really can't predict what will happen in 10 years. We can only foresee the next few years, and we must realize that the situation in 10 years is full of great uncertainty.

How Hinton views the relationship between AI and humans

The above are Hinton's latest interview and views.

But if we connect the dots, we can still find that the old man Hinton's view of the relationship between AI and humans has been constantly changing as AI progresses and its capabilities emerge.

If we go back 10 years, his view of AI was actually not as radical as it is today.

As one of the most important promoters of the deep - learning revolution, he long believed that neural networks could simulate the working mode of the human brain, but at that time, AI was still just a tool.

They could recognize images, understand speech, and assist doctors in diagnosing diseases, but they were still an extension of human capabilities.

Humans were responsible for setting goals, and machines were responsible for performing tasks. The relationship between them was similar to that of the steam engine, electricity, and the Internet.

The real turning point occurred after the emergence of ChatGPT.

As large - scale models demonstrated unprecedented language understanding, reasoning, and knowledge transfer capabilities, Hinton began to realize one thing:

We may be creating the first being in history whose intelligence level can continuously approach or even surpass that of humans.

After leaving Google in 2023, he almost devoted all his energy to discussions about AI risks.

During that time, his most commonly used metaphor was a "baby tiger".

It's like a cute little tiger, but unless you're very sure it won't want to harm you when it grows up, you should be worried.

But at this stage, Hinton's focus was still on how to control and constrain AI.

He even specifically criticized several AI giants for not paying enough attention to AI safety.

In Hinton's analogy, the concept of mother and baby began to appear. At first, humans were the mothers, and they needed to provide value and ethical guidance to AI and create safe AI.

He called on more countries and the forces of industry, academia, and research to join in. Don't just compete blindly, or the consequences will be unimaginable...

But last year, Hinton's analogy underwent a qualitative change -

AI is the stronger mother, and humans are the weak babies.

The baby is not smarter than the mother and doesn't have control, but the mother will actively take care of and protect the baby because she has maternal instincts and corresponding reward mechanisms.

Simply put, the old man believes that the future super - intelligence can no longer be controlled, and we can only hope that it will be kind to humans...

Just like a mother treats her baby.

Now, Hinton has made another astonishing statement:

AI already has consciousness, and humans will no longer be the only intelligent life.

The question is, if AI already has consciousness, why would it treat humans as its own babies?

A mother protects her own baby... right?

Full interview video: https://www.youtube.com/watch?v=p7t1Q_p2gZs&list=PLADd6sStSis77HKfbf4KCY6SvthfxeUgn&index=1

This article is from the WeChat official account "QbitAI", author: Tingyu. Republished by 36Kr with permission.