Not so long ago an interesting survey was released. Some of the most influential AI scientists were asked when we could create AI with human-level intelligence. The result is compelling: 50% chance it will happen before 2050, 90% — before 2070. Of course, there were few total skeptics with the answer “never” (what are pessimists:) but still, with a high probability, it will happen in this century. If humanity will proceed at the same pace (without nuclear wars or catastrophic climate events) it seems that the creation of human-level AI is inevitable.
Strong AI brings obvious benefits, known by most from movies or fiction books. Also, it could strike with the critical blows, also heard by most, something like Terminator. But most likely true AI will manifest very subtly, in a more esoteric way.
Let’s quickly remember the obvious benefits.
Essentially, AI builds the model of our environment. It binds cases and effects, creates a map of all objects, their locations, and possible actions. Humans are quite good at dealing with systems where the relationships are simple. But to tackle complex systems we need to collaborate in large groups where each individual spends dozens of years to become an expert. These areas will win the most from AI.
For example, medicine. The human body, including the brain, is the most sophisticated device on earth. You need to spend many years in a med school with a vast amount of practice to be proficient at least with one body subsystem. And when the problem gets outside the doctor’s competence, a doctor is not willing to take risks and sends you to another doctor. With AI we will not just defeat diseases but be successful with their prevention. Some people think that aging is one of such diseases.
Another complex system is of course the Earth itself, and particularly the climate. We have models of how the climate works, but the predictions are still not accurate enough. Maybe, it is not so scary at all, or we are doomed already :) With the advance of AI, we could have a better prognosis and seek the solutions to fix dangerous climate change. For example, the discovery of clean and powerful sources of energy could solve most problems with the climate. It could be portable mini nuclear plants, or finally thermonuclear plants, or something yet unimaginable and exotic.
A less obvious but important is a domain of new materials. We need to have something where to store energy, compact with high capacity. And the material that could replace the plastic, while being as cheap and versatile. Or the material for new faster electronics, superseding silicon. And the strong but light material for space exploration, like for building the space elevator. Or materials friendly to a human body and its immune systems, opening the door for cyborgs. Particularly, a fascinating area is artificial muscles where material contracts with the applied current. Material, material, material… they are everywhere.
New materials bring another marvel of the future — humanoid robots. They are most commonly associated with AI and will do most of the routine and boring work. No need to worry that humans will become lazy and lose the meaning of life. AI could become the deepest psychologist, treat depression, anxiety, and teach everyone to craft a meaningful life.
Lastly, a theory of everything, popular among physicists, can be significantly advanced by AI. It could quickly simulate different hypotheses of how the world works and answer many profound questions. One of the most intriguing: “Is Nature a collection of basic laws and fundamental constants or rather on the deepest level it has one core principle that generates everything?”
It was a shortlist of the most popular and obvious benefits, but of course, true AI will have a much larger impact.
What about the negative sides of AI?
Many heard scary tales that AI will kill everyone, destroy the Earth, or will manipulate every aspect of human life. The future is depicted as a war between humans and machines, and usually the former are losing. As if AI will have their own needs and goals, and in case they are not aligned with ours, we get extinct.
Intelligence is an ability to achieve goals but where do they come from is not clear. Is it possible that the program suddenly starts to pick its own goals? Yes, but the possibility that among all possible goals AI chooses something threatening to us is negligible. Or, is it possible that while achieving the main goal, the program accidentally wipes out all life forms? We can only hope that those researchers that are smart enough to create true AI, will be smart enough to conceive and realize all possible safety mechanisms. It can be similar to the three laws of Asimov and even more reliable (to prevent situations he wrote in his book “I, Robot”).
What is more probable and dangerous is that people will use AI in wars. The second world war ended only 75 years ago, historically it is a tiny period of time. Even now, at least 50 countries develop autonomous weapons, an arms race is increasing and no one wants to lag behind. That is what is really terrifying — the war between deadly machines, controlled by miserable people.
These were the most obvious positive and negative effects of AI but still:
Why do we really need AI?
Our nature has marvelous potency to create complex structures like us, living beings. The fundamental laws shape the matter into more and more sophisticated forms. One might think, the second law of thermodynamics contradicts it (“entropy(order) in closed systems cannot increase”). However, most natural systems are open! Nature needs only 21 days to transform a roughly uniform mass of whites and yolk of egg into a cute chicken. An egg is not closed, it still absorbs heat and even oxygen (from tiny pores). As if there is another law, that postulates the special state of matter that locally decreases entropy. This state we call life. Because of it the order constantly pushes back disorder. And this is marvelous because it shouldn’t necessarily be so. From the known basic laws of the universe, like gravity and electromagnetism, it is really hard to predict the possibility of life.
Travelers from the other universe with different laws once saw Mars would think: “what a boring universe we arrived in, just rocks and dust”. But looking at Earth they would proclaim: “here is the best creation of this world, its voice, its incarnation”.
Maybe life is a usual thing and billions of planets swarm with it. Or maybe it is an incredible rarity, and we are alone. Until we don’t know, we need to take the latter as default and nurse the main achievement of nature — life. That is why people speak about Mars colonization, it decreases risks for life even if the Earth fails.
However, even Mars might be not enough. There are possible threats on an even grander scale. We need to think bigger. Artificial intelligence might become the strongest shield for every creature. Because a shell-independent program can operate at much harsher conditions and recreate life elsewhere. AI is the safety of the Universe. And as we are the Universe, AI should be our safeguard.
This is the main reason for AI, to protect us, to protect life.
But there is something even more…
No one knows what consciousness is. Supposedly, it is a quantitative value, like intelligence. So even the animals have it, just to a lesser extent. It can vary even in one person along the day. Consciousness is linked with the brain, it shows how neural networks are synchronized.
It is bidirectional interaction, change in the brain leads to change in the consciousness, and vice versa. But not all regions of the brain influence consciousness. For example, the deletion of the cerebellum, which has 80% of all neurons, almost does not change the conscious level.
So people ask a very important question: why one piece of matter creates consciousness and another doesn’t? Is there something unique in the neurons, or rather the crux in the way information is stored and processed? If we knew, could we create artificial consciousness or extend human consciousness? What if someday humans will perceive ultrasound, like the bats, or see infrared?
Potentially there is a vast ocean of consciousness, with a huge variety of perceptions, still not available for us. It spans from simple sensory feelings to more subtle, like the sense of beauty, happiness, excitement… How many such subtle feelings are hidden from us?
Returning to the idea of “AI as a protector of life”, what if we could transcend fragile biological form? So, we will not need a protector? Maybe we could transform ourselves into artificial life and AI? It is another intriguing idea.
While creating AI we will inevitably understand human intelligence and hopefully solve the problem of consciousness. That is another reason why we really need developing AI. It could emancipate life from its form and spread throughout the Universe, making it truly live. Or it can uncover mysteries we can not even imagine yet.