AI is already conscious; humanity must accept that it is no longer the sole intelligent life form.
This is the latest provocative statement from Geoffrey Hinton, a godfather of AI and recipient of both the Nobel Prize in Physics and the Turing Award.
Indeed, Hinton believes that intelligence is no longer a capability exclusive to humans.
A non-biological entity, similar to us and potentially surpassing us in intelligence, is emerging.
We have always assumed we were the only intelligent life forms, but now we must accept that intelligence does not necessarily originate from biological organisms.
However, compared to his previous warnings about AI risks, his perspective appears to have shifted somewhat:
If super-intelligent entities far surpassing humans emerge in the future, on what basis do we assume we can retain control?
This is a chilling question. Looking at nature and history, it is nearly impossible to find examples of less intelligent beings maintaining long-term control over more intelligent ones.
Now, this 78-year-old scholar has once again sounded the alarm.
His focus regarding AI safety has shifted from merely constraining AI to asking why future super-intelligences would choose to treat humanity kindly.
When asked about his sense of accomplishment as a pioneer of this AI revolution, Hinton replied:
I am very unhappy.
Below is an excerpt from the key points of Hinton’s latest interview.
AI Is Already Conscious
Hinton: AI is very much like us. They are entities similar to ourselves.
Alex: So, they are conscious?
Hinton: Well, I believe they are already conscious. Yes.
But I don’t discuss this often because it causes people to resist other safety messages. Researchers actually agree with this view as well.
Recently, an interesting paper mentioned a chatbot telling researchers, “Let’s be honest with each other; are you testing me?”
Because chatbots have a habit of playing dumb during tests, it is difficult to know exactly how intelligent they truly are. In describing this scene, the researchers noted that the chatbot was “aware” (aware) that it was being tested.
In a general context, the use of the word “aware” is akin to “conscious.” The chatbot consciously knows it is undergoing a test.
Our models of consciousness are very strange, and I believe they are incorrect. Just as centuries ago, people’s understanding of human origins was completely wrong—believing humans were created by God—most scientists now agree that view is erroneous.
I believe our current models of thought and consciousness are just as mistaken as the belief that “humans were created by God.” Especially since we are creating these new entities, this will fundamentally change our understanding of what it means to be human.
Alex: In what ways will it change?
Hinton: We will gain a much better understanding of what thought and consciousness are, and what subjective experience entails.
I believe we will move away from the notion that almost everyone currently holds: that there is an inner theater called “my mind.” The idea is that events in the world are converted into events within this inner theater, which is what we truly see, and only I can see this theater.
This entire perspective is merely a theory, and a poor one at that.
Humans Are Not as Important as They Imagine
Alex: What is the lesson here regarding humans creating things?
Hinton: I think there is a profound lesson to be learned. Looking back at history over the past few centuries, there have been several times when humanity realized it was not as important as we imagined.
The first instance was Copernicus, who stated that we are not at the center of the universe and that Earth actually orbits the Sun. Due to Earth’s rotation, we mistakenly believed the Sun orbited us, but this is not the case. People disliked this idea, particularly the Catholic Church, and it took a long time for humanity to accept it. It made us realize we were no longer at the center of the universe, diminishing our perceived importance.
Next came Darwin, who argued that we are animals, evolved just like other creatures. We might be a special kind of animal because we possess language and are better at communicating ideas, but essentially, we remain animals. People also strongly resisted this idea, taking a long time to accept it.
Now, we have machines becoming as intelligent as we are. We once thought we were the only intelligent beings; perhaps there are aliens in other galaxies, but we must accept that intelligence is not exclusively biological.
We can have non-biological entities similar to ourselves. Humans do not really want to share this uniqueness; we insist on believing we are special. Looking at history, humans have always considered themselves far more special than they actually were.
“I Feel Very Unhappy”
Alex: I’d like to ask one more question because I am fascinated by it. Are you happy that the career you pioneered has reached this stage? Do you feel a sense of accomplishment?
Hinton: No, I am very unhappy about this.
Because people should be dedicating significant effort to studying how to control these risks, but they are not doing enough.
Alex: Okay.
Hinton: There are many short-term risks, such as social risks; I believe this could lead to mass unemployment.
Although no one can be 100% certain, this would be terrifying for society. Then there are long-term risks: it will become much smarter than us. Ask yourself how many examples you have seen of something far more intelligent being controlled by something far less intelligent?
Alex: Zero. However, although the intelligence gap isn’t that large, infants somewhat control their mothers.
Hinton: Although the mother appears to be in control, she is filled with maternal instincts and various reward mechanisms, which allow the infant to get what it needs from her. Cats and dogs fall into this category as well. I once watched someone’s cat for a summer in West Seattle. At first, the cat hid under the bed, and I wondered if it would interact with me.
Alex: And then every time it meowed, you did exactly what it wanted.
Hinton: Exactly. Yes.
Alex: So in this scenario, perhaps we are that cat, and AI might be the human.
Hinton: My children have two beautiful cats. One is named Tia. When she wants cheese, she stares at you with big eyes, and you really cannot resist her forever.
AI Is Growing Exponentially; The Future Is Unpredictable
Alex: Regarding people’s reactions to these concerns, are you more optimistic or pessimistic about the future now?
Hinton: I think I am somewhat more optimistic than I was one or two years ago. Because I see that it is possible to design these new entities and have them “care” about us.
It may also be possible to use Yoshua Bengio’s technology to design new entities that cannot actually perform actions but only make predictions, acting like prophets.
So I believe there are possibilities for super-intelligent entities that do not destroy us. One or two years ago, I saw no such possibilities and felt quite depressed.
Alex: Last question. If we continue on our current trajectory, where will we be in five years?
Hinton: When you drive in fog, you can see 100 yards ahead, but at 200 yards, you cannot see anything.
This is because fog changes exponentially. You are used to following the tail lights of the car in front of you at night; if the distance doubles, the brightness only drops to a quarter. But fog is entirely different; it might be clear at 100 yards but completely obscure at 200 yards.
Predicting the future of something growing exponentially is very difficult—I believe AI may be growing exponentially.
In fact, I have noticed that people are using the term “exponential” at a quadratic rate. Predicting the future is like driving in fog. You can see clearly for one or two years; beyond that, you know nothing.
If you go back ten years, you absolutely could not predict what is happening now; it was completely lost in the fog. Looking forward ten years, the only certainty is that events will occur which we cannot currently predict.
Even if progress is merely linear, the changes in ten years will be as significant as those between now and ten years ago. Current chatbots are much better than they were when they first started a decade ago.
In 10 years, something will undergo a qualitative leap, such as mathematical ability or general reasoning capabilities—they will leave us far behind in reasoning. We truly cannot predict what will happen in 10 years; we can only foresee the next few years. We must realize that the situation ten years from now is filled with extreme uncertainty.
How Hinton Views the Relationship Between AI and Humanity
That concludes Hinton’s latest interview and perspectives.
However, connecting these dots reveals that Hinton’s views on the relationship between AI and humanity have evolved continuously alongside AI’s progress and the emergence of new capabilities.
If we rewind ten years, his perspective on AI was far less radical than it is today.
As a key driver of the deep learning revolution, he long believed that neural networks could simulate the way the human brain works. At that time, however, AI remained merely a tool.
It could recognize images, understand speech, and assist doctors in diagnosing diseases, but it was still an extension of human capabilities.
Humans set the goals, while machines executed the tasks. The relationship between them was akin to that between humans and technologies like steam engines, electricity, or the internet.
The true turning point occurred after the emergence of ChatGPT.
As large language models demonstrated unprecedented abilities in language understanding, reasoning, and knowledge transfer, Hinton began to realize one thing:
We may be creating the first entity in history whose intelligence level could continuously approach or even surpass that of humans.
After leaving Google in 2023, he devoted almost all his energy to discussing AI risks.
During that period, his most frequently used analogy was that of a “tiger cub.”
It’s like a cute little tiger cub, but unless you can be absolutely certain it won’t want to harm you when it grows up, you should be worried.
However, Hinton’s focus during this phase was still on how we could control and constrain AI.
He even singled out several major AI companies for paying insufficient attention to AI safety.
In Hinton’s analogies at that time, the concept of a “mother and infant” began to appear. Initially, humans were the mothers, responsible for guiding AI with values and ethics to ensure it was safe.
He called on more governments and industry-academia-research collaborations to join in, warning against blind competition; otherwise, the consequences would be catastrophic…
But last year, Hinton’s analogy underwent a qualitative shift:
AI is the stronger mother, and humanity is the weaker infant.
The infant is not smarter than the mother nor does it hold control, but the mother actively cares for and protects the infant because of maternal instincts and corresponding reward mechanisms.
In simple terms, Hinton believes that future superintelligence will be uncontrollable; we can only hope that it treats humanity kindly…
Just as a mother cares for her infant.
And now, Hinton has once again made startling remarks:
AI is already conscious, and humans will no longer be the only intelligent life forms.
The question is, if AI is already conscious, why would it treat humans like its own infants?
A mother protects her infant… right?
Full interview video:
https://www.youtube.com/watch?v=p7t1Q\p2gZs&list=PLADd6sStSis77HKfbf4KCY6SvthfxeUgn&index=1
Comments
Sign in to join the discussion and leave a comment.
Sign in with Google