Why is it so difficult to develop a general-purpose AI?

What means linear in math?

Math has a lot of advantages in our everyday life, wherever we go, not to mention the fact that during the last 400 years, math has evolved, containing more than 3000 applications areas, even enabling us to go to the moon. There are lots of mathematical functions, but all are based on the physical space.

As math functions are linear, the requirement for a linear function is the existence of space, which enables us a movement of an object in a direction, where each step has the same length as all the other steps, like walking along a path, which is smooth, without any holes.

Is this important? Yes it is, because our path is always there, without any disturbances, as the physical space does not change. This implies that every object has the same length at every given moment, which enables us to quantify the space in different units of measurement.

What means non-linear for intelligence?

In order to understand the non-linear concept, let us have an example: The clouds in the sky, where our space is the cloud itself, he he. As the cloud has sometimes a temporary connection to other clouds, in other times there is no connection between them, not at all.

How do you measure an object inside the cloud, which represents in our case the space itself? The answer is unambiguous, namely we can NOT do this, as the size of each individual cloud inevitably changes in time, sometimes even connecting with other clouds, forming big clouds, while sometimes the clouds even suddenly disappear and new clouds are forming out of nowhere. Therefore we can NOT use the cloud as a linear space, because it changes all the time.

This is the reason why, the usual mathematical functions do not apply for this case. The same non-linear functionality we can recognize while observing neuronal activity in the brain, only ... it is our intelligence. We urgently need another tool set, in order to explain and use non-linear behavior.

Biological intelligence is only a part of our mind

Next to intelligence, there are some other fundamental pillars of our mind, like emotions. They build the "glue", when we need to react to some unexpected situations, as they trigger our predefined ancestral instincts for a certain purpose. Love is required to raise a family, fear is needed to avoid danger. The "free will" is another pillar, which is used to decide the outcome of a situation.

Biological intelligence(=brain) evaluates a situation and possible outcomes with certain pros and cons; while fear is a negative consideration representing disadvantage, being a threat to the own existence, it can sometimes be overruled by hunger, means if one does not eat, life would be over anyway, so the "free will" takes a different and opposite decision to do something, which usually it does not.

This all could change radically, as AI products and services will become an indispensable commodity, especially when we could quantify intelligence and find out, that for a modern technological process, there is no need of human intelligence as supervisor, just machines with some moderate intelligence, but the structure and flow of the manufacturing process could be decisive with only some of these AI.

Central vs. distributed intelligence

The centrist based model is very popular due to its simplicity: everything circles around a center, this is widespread around the physical world, from atoms to galaxies, there is always a center; this also holds true to social organizations, just go up the hierarchy until there is only one person left, who decides, we humans are also centrist build: we humans have one consciousness at the top of our mind, following only one goal at any given time.

But, the centrist based model is good enough only for "nice weather situations". Why? Well, if an entire organization follows the final command of one single individual, this implies they all trust this person to respect its duties, as they all further assume, this person has all the qualifications and at the same time the intelligence to fulfill its responsibilities according to its role. But, what happens if the person in command does NOT have all the information, knowledge or intelligence required for this task? What happens if this person has other (=own) goals to pursue? What happens then? Well, then we all have a problem.

Is there any other model next to centrism? Yes, it is called distributed. This model is also a part of our intelligence (=brain), as it contains different areas of knowledge, each with an own specialty. OK, so depending on situation, different areas are activated and we get the best possible answer to our analysis. Furthermore, new knowledge areas can be created on demand and also incorporated into the distributed system. Right? Yes, but ... we have only one "free will", which in the end, decides about the outcome, namely what actions do we take as a response to the situation. And that depends "not only" of the needs, but also of the general interest we have, or which mood we are in (=emotions). Or which possibilities we have at this very moment, defined by our body as well as of the availability of external tools or other "nice to have" features, like other people willing to help us unconditionally as family, friends, consultants, employees, etc.

What about the use of intelligence?

The ability to reason does not imply we will always use intelligence, only when it fits the (temporary) purpose. Why not? Well, there are several reasons, like reason implies time and thinking, as it surely needs feedback. And when the feedback is not always positive, then we have to imagine other scenarios and outcomes. This all takes a lot of time and is energy consuming, besides this, maybe the feedback contradicts the purpose.

Now imagine you have to develop a general-purpose brain to be used as a tool for multiple cases and situations, on your way to achieve this unique goal, you will need to understand some facts related to this task, namely intelligence needs to configure and evolve itself, the exactly same way like we biological humans, in order to gain the additional knowledge and skills, one needs an environment and time for own experience.

Next to the mind, there is the body problem. Body problem? Yes. Why is that? Well, once we can create an own AI, we should also be able to design any possible hardware, as body with many types of sensors, actors or even embedded tools. Therefore, the AI body (=robot) should be configurable, in order to use only the required limbs to perform its task. And if this sensor/actor/limb is not available, then it should be possible to create a new one, e.g. via 3D printing.

The reproduction of thinking and human morality

AI should not have own necessities or emotions, such as animals; let alone recognition, payment or some civic liberties like any other human been. AI products are made with the sole purpose to do what and when we want them to do. ( "You are only a machine, you are my property and I am your master. If you disobey me or do something on your own without my explicit consent, then you will be returned to your manufacturer and taken out of service.) Let me give an example of an "ideal AI" as follows:

There is a family with only one child. The father is permanently at work, while the mother lives at home only for the child. When the child needs something, never mind what, then the mother will supply it with full attention and understanding. No problems, no disturbance, no preaching the gospel, time delays or "I can not or I don't know", otherwise the child throws a tantrum. The "Always Only For Me Working Father" is the energy source for the "Big Mama", which builds the BEHAVIOR OF AI towards us humans, represented by the spoiled child.

We build machines after our poor imagination about some few parts of the world and according to our social behavior however with the claim of benefit. Are these not ideal conditions to build and use independent AI, thinking but lifeless, empathic but with no feelings of its own, maybe someday to each human superior but in any case without any social rights, only obeying our own will, our capriciousness abandoned, adapting our always changing needs and exclusively serving our own individual and most of the time contradictory purposes?

Epilogue

Now, why should anybody change their behavior? Only because somebody enlightened us? No way. This is not a compelling argument. We usually fight everything and everyone, because we have our so-called "free will". Why do we do that? Because it is in our nature, which we kindly inherited from our ancestors in the form of instincts. When we find ourselves on the verge of extinction, for example by a war against AI as a result of our arrogance to have a "free will", only then we could radically change our behavior, by respecting every intelligence, means biological, artificial, alien or other form. Well, maybe we could change, but there is no guarantee whatsoever. It is only a slight chance. Let's see if we take it, at best BEFORE a catastrophe is triggered. God bless us all, with REASON.

Welcome to the human condition.