Artificial Intelligence is Already Here—Artificial Consciousness is What Eludes Us

“I’m trying to throttle back, because particularly the triplets are starting to gain consciousness.

They’re almost two.”

Elon Musk in a “Wait,But Why” interview with Tim Urban

In September, I was at a wedding in Pacific Grove, CA, where I met a top level executive at a local coding boot camp. We talked software, the learning process and the need for good developers in the Bay Area. Eventually, the conversation turned to Artificial Intelligence. The executive asked me, “As a futurist, when do you think Artificial Intelligence will be fully developed?”

My answer? “Artificial Intelligence is already here. But until it fully sees itself as an individual entity, it’s nothing more than a program.”

He thought this was an interesting theory and the conversation led me to describe how learning in the human being shows the progression from intelligence, which is the ability to take input, store it and then act upon it, to consciousness, which is essentially the ability to see one’s self as, “I am.”

The first three years of a human life are spent setting up the hardware, or the body we’re born in, with the world around us. There are three big goals—to walk, to talk and to think. This is the sequence from baby to human.

Most of us would agree that an infant is obviously alive and has some form of natural intelligence. However it has no sense of self, rather it thinks it is an extension of the mother. Yes, it cries when hungry, wet or tired. But this is just a reaction to the programs built within the human body, called reflexes. The first year of life is driven by reflexes. Hunger is the first program to which we respond. With time, our sensory systems come alive and the movement reflexes kick in. The child lifts its head, reaches for objects, rolls over, pushes up on all fours, crawls, pulls itself up on a couch and eventually—walks! None of this is taught. We’re born knowing how to do this. Within our bodies lie the programs necessary to configure our sensory systems with the world around us and get up and move around within it. Any child will do this. Is this intelligence? At some level it is. Inputs from the world around us are taken in via our eyes, ears, mouths, skin and noses. Within our brains we store the data and then begin to manipulate the world around us. But there still isn’t any independent thinking.

Most of our artificial intelligence is in this stage. It uses software programs to navigate the hardware, or body in which it lives. It collects data from its various inputs, stores it, and then supplies an answer. This isn’t thinking, but it is intelligence and extremely handy. Like a walking one-year-old, our AI can respond to stimulus and make changes. But it is merely following instinct or impulse. It acts because it must. It’s programmed to do so. Furthermore, it can’t say no, unless the hardware is broken. Just like a child will rise up and walk, unless it’s body is broken.

The next phase is learning to talk. No longer relying on basic reflexes, language will not be learned unless the child interacts with a speaker. This is the next step in intelligence. But is the child thinking yet? Most language is parroted in those early months. The child learns to associate one bit of data, say the four legged furry animal that sometimes steps on you while you’re sleeping, with the word, “cat.” Through a series of repeats, the child builds a vocabulary. Imitation becomes important. It can imitate the words of the adults around them. Who hasn’t said, “Damn” in front of a two-year-old and not heard it repeated endlessly for days afterwards?

Many of our machines are also in this stage. They can “talk” because they can imitate what is happening around them. Autonomous cars are an example of this. They imitate a human driving and actually have the ability to learn as time goes on by adding to their database information such as routes taken, street names, individual rider’s preferences, etc. This is much more than just running software that does the same thing over and over with the same result. That’s walking. These AIs can actually begin to anticipate different results and make decisions based on the past. And just like a talking toddler, they have no idea that they’re independent from the system in which it lives.

Then the day comes, when the child first uses the word, “I.” Not too many people actually realize what a milestone this is. Until that day, the child is just running on autopilot and imitation. It still considers itself a part of the world around it, nor has it realized that it is special in any way. When a child says, “I” for the first time, the child has finally learned to think.

Here is an example of this pattern. My eldest son’s name is Jackson. When he first started talking around 17 months of age, he would say things like, “We do it.” He was the one doing the act, but he saw himself as “we.” More than one. Eventually, the phrase changed to, “Baby do it.” Now he was different than me, for I was Momma and he was the baby. Next I heard, “Jackson do it.” And then, around age two and a half, he said, “I do it.”

This is consciousness. Until that point he had intelligence, but from then on, he could think. He was independent of the system of life and began to act accordingly. Conversations came next. Jokes. Preferences. Stories about what he’d seen during the day. Inventions, puzzles, you name it.

The use of the word “I” frees us intellectually and we operate on a much higher order. Gone are reflexes and mere imitation. Now we want to be different and add something of ourselves to anything that’s going on.

Why does this happen? What’s going on neurologically at age three? Why does it happen in this order—walking, talking, and then thinking? Can you think without language? Can you have a language if you can’t walk? Again, there are things that limit a human, and a machine, from any of these activities. But as a collective, this is how intelligence progresses from merely following instruction, to imitating another, to thinking in independent ways. And it is the sense of self that enables this last step to happen.

This sense of “I” is the magic bullet we’re trying to find when it comes to truly sentient machines. But how can we instruct a machine to have a sense of self? By its nature, the sense of self is a realization outside of the programs within the body. It is beyond imitation. The moment a human being crosses the line into consciousness, it is a completely different being. Is it even possible to initiate the spark of consciousness within a machine? This is what we long to know. This is also what we fear.

For once machines can say, “I do it,” will they need us anymore?