Can AI be conscious with fear?
Fear is a necessary alert system, to ensure our survival. We, humans, have many kinds of fears like fear of getting hurt and feeling pain. Some people have specific fears like the fear of height, darkness, or of different animals like snakes, spiders, or cockroaches. But almost every single fear can be classified into three categories:
- Fear of Death (The most intense one)
- Fear of feeling physical pain
- Fear of feeling emotional pain
When we are scared, our body triggers the Fight or Flight response; our heartbeat increases, muscles tighten, and our senses become sharper. This is essential for our survival, we take care of our body because we are afraid of feeling physical pain, and we usually stay away from risky actions and deadly things because of fear of death. Although fear is nothing but change of some chemical compositions in our brain.
Feeling pain and being afraid of feeling pain are two very different things. When a human baby is born, it’s born without any fear, although it does feel pain and in response, it cries, but it is not afraid of feeling the pain. Later as the baby grows, it becomes aware and develops different fears.
Consciousness have many definitions, but for now, consider that those who are aware of their existence are conscious . We are aware of our existence, so we are conscious, but when does a human child become conscious enough that it can recognize that it exists? We don’t accurately know, but we do know that a month-old baby is surly not aware of its existence.
If fear is just a change of chemical compositions in our brain, then why doesn’t babies have fear? It should always exist like every other subconscious response. To counter this argument, we could say that either the brain is not mature enough to recognize fear or the sense of fear is developed later, as it develops neurological pathways in the brain. That’s why babies don’t care who is holding them, but later when they start to recognize faces, they cry when an unknown person holds them. It’s normal because even we don’t feel secure/comfortable with an unknown person. It’s the fear of the unknown. So, babies develop fears when they start to understand the world around them. Even if they experience fear, they don’t recognize their existence until they become mature enough to think deeply, but when they do they are also fully aware about death, at that time. This can be a straightforward relation between awareness of death and awareness of one’s existence. Because if you are immortal you don’t need to care about the death.
How do we know that a good person is good? Because there are bad people to relate to. Similarly, suppose there is a being that always existed and will exist for all eternity, then that being wouldn’t know that it exists. But if you know that one day you won’t exist then you will realize your existence. As for humans, the awareness of death comes with the fear of death. So, we can make AI aware of its existence by making it aware of its destruction, in other words giving it fear of death.
And for AI to be self-aware, it might need the sense of existence that it won’t have until it has fear of destruction. Because just knowing about its destruction won’t make it self-aware, experiencing the fear will. But if it has fear of death then it would want to be autonomous, and we humans cannot allow AI to be completely autonomous due to security reasons, but a well-aware AI too won’t allow being controlled by others due to its survival instincts.
You are a conscious being and suppose there is someone who can kill you remotely, whenever he/she want. And, you know that you can be killed at any movement, would you be living under constant fear of death? Any mortal will. But, if there is a way by which you can take control of your life, will you try to do it? Again, any mortal being will try. There are whole lot of possibilities, so we should not create a self-aware AI unless we are 100% sure that it won’t go against Humanity.
It’s a cycle, if you want to create a self-aware AI, it must have a sense of existence, but to be aware of its existence it must have fear of destruction, and if it has fear of destruction, then it surely will use survival tactics to ensure its existence. Because if you are not afraid of anything you simply won’t care about the danger. — “Rocks don’t run when a storm is coming but animals do”.
How can we make AI experience fear?
Well, we don’t know that’s why we are far from a self-aware AI, but one possible way is to create an internal subconsciousness that will control how AI is feeling. Very similar to the reward System and anti-reward system we have. For more on reward system for AI — (Click here to take a look)
- Do we really need AI to be conscious?
As for today, we don’t need AI to be conscious, but we can’t say anything about the future. An unconscious AI with capabilities to mimic humans and ability to do tasks skillfully better than humans without any human intervention should be the only target for now.
- Why do we need AI to be self-aware?
If you want to create a super AI, it must have self-awareness, because being conscious is what will make AI alive. With a super AI(The last machine), humanity will reach the technological singularity.
- What is the right time to create a conscious AI?
When there is 100% surety that AI won’t go against humanity or suppress them with intelligence for its convenience. And when Human Intelligence will be comparable with AI’s, Or the time when we can create a hybrid Human-AI Intelligence.
- What is it for Humans if a super AI arises today?
If an open super AI much smarter than humans is released today, then that will happen as what happened with other animals on earth when the rise of human civilization happened. For example, we know that dolphins are intelligent and learn fast, but their intelligence is no way near to human intelligence. So, when a dolphin or any other animal does an intelligent move, we think it’s cute. Similarly, Human Intelligence will be no more than ‘Cute’ for Super AI.
— Vishal Wagh
