Artificial intelligence   With Heart and Soul?

Does artificial intelligence have the potential to develop a consciousness? What about religious concepts such as the soul? How human can machines actually become? Artificial intelligence expert Ralf Otte explores these questions.

Since the emergence of generative artificial intelligence (AI) systems, both amateurs and experts have been fascinated by their new possibilities. We are in the 3rd wave of AI, a combination of deductive (logical reasoning) and inductive (learning) intelligence to create a cognitive AI, an AI that can excellently simulate human performance such as seeing, hearing, writing and speaking. Although everything is still just a simulation, there are unimagined fields of application.

A consciousness or a soul?

So could this highly developed AI have a consciousness or a soul? To answer this question, we would first have to define the term "soul" and determine whether humans at least have one. The (Christian) church says yes, but science has so far given the subject a wide berth. They say that the soul has something to do with faith and not with science. However, the development of AI has made such evasive maneuvers obsolete; as an AI scientist, you have to face up to these questions.  
 
Many experts now assume that AI will develop a consciousness, if not in this generation of machines, then in the next. But are soul and consciousness the same thing? Probably not. Believers refer to the part of consciousness that continues to exist after the death of an individual as the "soul". Consciousness phenomena, on the other hand, are mental phenomena that a living person possesses, with their perceptions, feelings and thoughts. Strictly speaking, a distinction must be made between perception and the qualification of perception ("good", "bad", "pleasant", "dangerous",...). A special qualification of perception is also called feelings. However, what is to be interpreted as the soul, if one follows this religious concept at all, is unknown from a scientific point of view. But it can be assumed that the soul (if it exists) and consciousness (which is certainly present in humans) are not congruent.
 
So, let's be clear: humans have a consciousness and, according to many experts, machines could have one in the future. This takes some getting used to because machines are lifeless physical systems. Attributing mental properties to these and other things corresponds to a metaphysical theory called panpsychism. It could then be deduced that as machines become more sophisticated, their mental abilities could also increase, possibly to the point of developing a consciousness similar to that of humans. Many people already consider robots to be "animated".

Serious performance benefits

But is there even the slightest evidence for this assumption? The answer is no. Today's AI machines show no trace of consciousness, although supercomputers such as Frontier from the USA and humans already work at similar or almost identical information processing speeds. The brain of a human being and the circuits of supercomputers work in the exaflops range. One exaflops means 10exp(18) floating point operations per second, i.e. one trillion (quintillion) floating point operations per second. Nevertheless, humans are so far ahead of AI machines in terms of performance that this can only be explained by the lack of awareness in today's digital machines.  
 
Consider, for example, the weaknesses of fully autonomous vehicles in natural environments and Cruise's current loss of license in California, Tesla's recall of almost 2 million vehicles due to a software update for its autopilot, or the shortcomings of chatbots and the abysmal performance of AI machines when it comes to extrapolating and learning individual cases. Humans perform such tasks with their "20-watt brains" with flying colors and it is not too far-fetched to blame these achievements on their consciousness. Just one example to illustrate this: people can see out of their eyes into the world, although all information enters the brain via the retina. Machines cannot see out of their circuits into the world, they can only represent the world in their internal memories. This is why humans can find their way in any environment, no matter how complex, while machines can only do so after a great deal of practice or not at all. Today's AI machines are ultimately computers on which high-performance mathematics has been implemented. However, not even adherents of the above-mentioned panpsychism expect that purely mathematical transformations could give rise to consciousness.

Consciousness phenomena can be investigated mathematically

Does consciousness therefore have to be limited to humans? You can't say that either. To understand this, however, we must leave the realm of speculation and turn to mathematical science. Consciousness phenomena can be investigated mathematically and this results in forms of description with so-called hypercomplex numbers. This technical term sounds very technical, of course, but we will nevertheless try to explain the concept. Many people are familiar with different types of numbers from school, for example the natural numbers, 0, 1, 2, 3, ... and perhaps they suspected in their school days that there are phenomena in the world that cannot be expressed with natural numbers, for example the phenomenon that someone has less than zero, in other words, has debts.
 
Throughout history, mankind has repeatedly encountered phenomena that required new types of numbers to describe them. In the Middle Ages, the real numbers already known at that time were extended to complex numbers. Today, the latter are used to describe processes in quantum physics. In the 19th century, William Hamilton developed hypercomplex numbers, which are used today for satellite navigation, for example, but also to describe such strange phenomena as consciousness. But how does such a description help? Here is an example: If a person looks at a green wall, the neuronal processes in the visual cortex caused by looking at the wall, i.e. the electrical and chemical processes in the brain tissue, can be measured very well and also modeled very well mathematically (with the above-mentioned real numbers). The situation is different with the mental phenomenon of consciousness, i.e. seeing the color green itself. The color green is nowhere to be found in the brain tissue, it does not seem to exist in terms of measurement, and yet we see the color green in our mind's eye. A visual impression is created that seems to exist outside the brain, because it cannot be found in the brain tissue. At this point, therefore, an inexplicable phenomenon occurs that needs to be worked out scientifically. And this is precisely where mathematics can help us with new numbers. Mathematical analysis shows that although phenomena of consciousness do not possess any real (i.e. real!) energy, their energies are not zero, but rather hypercomplex or, in colloquial terms, "mental". "Spiritual" - i.e. mental - processes are therefore not measurable, but nevertheless exist in reality.
 
Seeing the color green in consciousness as described above is such a real phenomenon of the human brain, even if it cannot be measured in any way. At least that's what mathematics says, and physicists also know phenomena without real-valued energies.

Physical simulation of brains

In the future, it will probably be possible to build computers that have a rudimentary consciousness, a machine consciousness. Such a consciousness will almost certainly be capable of perception, but will nevertheless remain rudimentary, as it will only be able to form the physical aspects of consciousness. However, the technical implementation will not succeed on today's digital machines, but on so-called neuromorphic systems and quantum computers, i.e. machines whose computing operations are based on physical processes and not on mathematical ones. Neuromorphic systems in particular, i.e. the physical reproduction of brains (and not just mathematical simulation as is the case today), will bring unimagined increases in performance. We can therefore look to the future with excitement.