(author: Oleg Chervonogradsky )(translated by: Iana Chervonogradska)
Each science and its applications contain fundamental assumptions defining the main direction of researches and practice, which in their turn always appear to be controversial and become constrained because of these contradictions. The aim of thinking is to show these contradictions and handle these limitations proposing an alternative.
The concept of Artificial Intelligence is not an exception in this case.
Opening notes
Before getting into the essence of the article, it is important to make several observations regarding the philosophical premises, as the basis on which the critics of the concept of Artificial Intelligence and a suggestion of new approach will be presented.
- Intellect (thinking and its forms) will be considered from the dialectic point of view, where it is a part of a human as social-historical formation, therefore the problem of the intellect is identified only in a social-historical context.
- The main difference between human race and animals is that a human can change their activities while an animal is “programmed” for the cycle of repeating the same functions. The human activities are reflected in the mode of production – appropriation and treatment of nature. But the most important and defining feature of the human as a generic being is changing the mode of production, i.e. – development of mode of production.
- Human is a product of mode of production development. Human is a social-historical product, and therefore an intellect is a product of social-historical progression, in other words it is a product of mode of production development in a social-historical context. Human as a social creature is not reduced to physicality which only contains a human and expresses them. That is why the consideration of intelligence cannot be narrowed down to the bearer – i.e. by biological, chemical etc. functions of the brain in particular. Neurophysiology of the brain should be learned in order to understand how thinking is provided but thinking itself is not narrowed down to the neurophysiology of the brain.
- Intelligence should not be considered as some kind of essence coming out of nowhere, but as an essence historically formed by society and constantly changing in the process of mode of production development.
- It is unacceptable to consider the intelligence in isolation of a universal connection with other aspects of life (perception, sensitivity, mind, imagination) and a human as a social being in general. Otherwise an opening to metaphysical constructions is inevitable, which will turn out to be for sure (and it happened already) idealistic illusions in “explanation” of the essence of intelligence as a derivative from spirit, God etc., that is contradicting the scientific approach and crucially decreases the practical possibility of modelling of intelligence (thinking and its forms) in reality, creating blurry and mystical background (Plato Ideas), which are not of any practical value, scientific-theoretical or productive value.
- The criteria of the truth is a subject-practical activity.
The opening notes above are sufficient in order to start the review of this article. For closer look at dialectics the author of this article recommends referring to the work “Dialectical Logic, Essays on its History and Theory” by Iljenkov.
Artificial intelligence: critics.
In 1956 the term Artificial Intelligence was coined as a field of study and practice at Dartmouth Conference (McCarthy, 2006). Here is a brief extract from a preparatory work of conference organization:
“The study is to proceed on the basis of the conjecture that every aspect of learning or any other feature of intelligence can be so precisely described that a machine can be made to simulate it. (McCarthy, et.al. 1955)”
This statement is a successful description of a conceptual approach to the problem of creation/imitation of intelligence (thinking and its forms), which was retained and still holds true today. And here is why.
Let’s consider the approach above in details.
It was proposed to found a study, aimed at “a precise description” of “an aspect of learning” or “any other feature of intelligence”, which can be simulated by “a machine”. Here we can obviously and clearly see a main presupposition of the approach, e.i.: to create a sufficiently precise description of a feature/aspect of intelligence learning, which is suitable for simulation by a machine, which means that simulation possibilities of the machine determine the mode of description of features/aspects of intelligence learning. Based on these suppositions it appears that computational capabilities of the machine determine the mode of creation/simulation of intelligence, so the bench-mark of the concept of Artificial Intelligence is not understanding of intelligence, but actual computational capabilities of the machine.
The unconditional industry standard of the calculating machine is a concept offered by Allan Turing named after its creator – Turing Machine (Turing, 1937). The point is that Turing’s machine is capable of performing only formal logical computation. Like in 1956 Turing’s machine is the only machine by means of which “each aspect or any other feature of intelligence”, its “precise description” is “simulated”. This means that at this moment the formal logical possibilities of Turing’s machine determine the mode of creation/simulation of intelligence – of thinking and its forms.
The contemporary researches of AI and moreover their practical applications are based on the formal logical computational possibilities of Turing’s machine. Though this approach was inevitable and necessary, still it is insufficient for creation of intelligence and its forms.
A famous Turing’s Test (Turing, 1950), the influence of which cannot be underestimated, is an example of the paradigm mentioned above to the approach of construction/simulation of intelligence. A formal logical handling of linguistic structures cannot be the criteria for determining successful creation of such a complicated and all round social historical formation as intelligence. This is a clear narrowing down of understanding of thinking to one of its features. We already mentioned this question in general philosophical notes.
On the other hand the Chinese room being an example of critics of Turing’s Test is putting in question the functionalism of Turing’s machine guided by the idea of biological naturalism, which in its turn appears to be another extreme example of narrowing down of intelligence to the neurophysiological features of the brain.
Artificial General Intelligence (AGI) is also worthy of mention. In spite of the fact that in AGI the concept of Artificial Intelligence is clearly pointed out as a constrained one and such that can solve restricted problems only (Goertzel, Penachin 2007), still in AGI the key problem is not brought into focus, but understanding of intelligence is just being reduced to the scale, the extreme points of which are formal logics and neurophysiology. And the integrative approach of AGI is cumulative approach in integration of those formal logical implementations of partial and constrained understanding of thinking and its forms, and furthermore taken out of the context of social historical progression, the product and the part of which is the intelligence.
Proceeding from the above: the contemporary paradigm of Artificial Intelligence in its researches and practical attempts on simulation and creation of intelligence (thinking and its forms) proceeds from the existing possibilities of calculating machines, which are formal logical and determine the ways of understanding of intelligence within this conceptual framework and actually limiting it.
This means that the implementation tool determines what needs to be done/created, i.e. the tool determines the task/problem statement, but not vice versa where the problem determines the choice and the necessary construction of the implementation tool.
In spite of the fact that this paradigm was historically inevitable, although quite useful, it is not sufficient and adequate for the establishment of the intelligence concept (thinking and its forms).
It should also be stated that this paradigm can be observed not only in the sphere of AI, but also in other spheres of theory and practice – sciences, which are also significantly constrained by this paradigm, or rather the researchers using this paradigm significantly constrain the possibilities of science development. But it’s not the subject matter of this article. Probably this topic will be covered more widely in another work. At this point after a short digression let’s come back to our question.
Controversy of AI concept
The paradigm mentioned above is only one of the constraints in modelling of human intelligence. The second significant constraint being the result of the first one is a false approach in understanding of intellect.
This approach is based on the direction, that the human intelligence is the feature common exclusively with an individual. Such an individualistic reductionism is the result of an instrumentalist reductionism.
This direction in the sphere of AI sets interpretations of the intelligence, which can be settled on the scale mentioned above – formal logics/neurophysiology.
The actual consideration of intelligence as a social phenomenon gives the right direction for the research in general, but they act within the framework of an instrumentalist paradigm, which again constrains the understanding of thinking and its forms. Moreover, these approaches ignore History as the main element in the formation of the adequate interpretation of intelligence, without mentioning the scientifically correct understanding of evolutionism and the necessity of revolutionism in the development (Bosenko 2001), i.e. the progression of History.
Intelligence (thinking and its forms) is arising and developing only in social historical progression of a human and yet is always manifested though an individual human being. But this manifestation should not mislead us to the definition that thinking is a feature of an individual human being within their limits.
When trying to simulate the intellect within the framework of the individualistic reductionism, a likeness of thinking, which can solve a wide variety of problems, e.i. being at least comparable to a human, is expected to achieve. But the point is that a human intelligence is a social historical product and that is why within the individualistic reductionism we fall into a contradiction where something that we did not expect to achieve is created. Maximum of what this approach can count on is creation of animal machines, which is actually taking place right now. But animals do not understand the Society/History – i.e. have no intelligence.
The intelligence being a part of social historical development cannot be created apart from this fundamental process. It can emerge only as a process of development within the social historical progression. But this will not be the process of a human being, but the process of a different formation, having its « society » and « history », an alternative one.
The predicate “artificial” assumes that artificial creation is surely a part of human bondage and cannot exist on its own without even mentioning the development beyond humanity. But the intelligence cannot depend on a human, it has to be independent/self-sufficient. Hence the term Artificial Intelligence is not an adequate term for the sphere of intelligence creation. That is why we introduce a term Alternative Intelligence.
Alternative Intelligence
When setting a goal to create an Alternative Intelligence we set a goal to create a material formation, which will develop as a social historical formation.
A social historical development of a human is obtained with the development of the mode of appropriation and treatment of nature – production. The main exponent of the mode of production is the instrument of production and the relation of the humankind regarding it. Based on the instrument of production development and the relation of the humankind regarding it, new features of a human arise and develop, amongst which is intelligence. That is why the only criteria of intelligence (thinking and its forms) that an artificial formation can have are:
- instrument of production
- relation regarding the instruments of production
By material formation, which can create instruments of production and build a collective attitude to it (i.e. practicality), we mean the organic formation having a sufficient structure and constitution of organs, which can perform these functions.
The creation and development of instruments of production always entails collective activity, which on the other hand produces the relation to the instrument of production – i.e. practicality of its usage, thereby the relationships within the community arise – social relationships. This is possible only within the framework of social development process, which is realized as a historical development.
The breaking point in the development of human kind was moving away from handicraft to agriculture. It was a turning point in the history of a humankind, which separated it from its reflective repetition “programmed” by nature, to the conscious activity, and hence, to the germ of intelligence and its development. Surely this was followed by numerous changes of ways human used and changed nature (for example industry), but the main element was the fact of changing of this activity and this was fulfilled in the development of instruments of production and the relation regarding them. That is why these two criteria are the key indicators of material formation having an intelligence, and hence the main criteria for the Alternative Intellect.
A lot of consequences arise from the above. We will consider only some of them.
- The development of instruments of production and occurrence of the new ones is the product of historical progression, conceived not only as an evolution (quantitative increasing of the same consistent quality), but also a revolution (qualitative change). i.e. a social historical development process through the development of instruments of production as an evolutional-revolutional process.
- The instruments of production are quite specific material objects, created to fulfill the human needs as a social being, and these needs are expressed by the way of using the instruments. This entails a need for material implementation of bearer of Alternative Intelligence, and it cannot be just a program without leaving behind some material artefacts of its activities, which in its turn are knowledge- and experience-holders, having a memory with a structure far exceeding the possibilities of the information structure coded in Turing’s Machine. The things created by a human are the only medium, where all knowledge and experience gathered within the social historical progression are saved and kept, and this progression is fulfilled in a thing. That is why in Alternative Intelligence it is fundamental to create the historical artefacts being the essence of experience and knowledge, the knowledge-holders. The quintessence of such artefacts is the instrument of production.
- The alteration of generations is one of the important attributes in development of society. The development of society would be impossible without human life constraints. The next generation uses and develops the heritage of the previous one. That is why one of the functional features of Alternative Intelligence should be generations and their alteration.
- This is the question of ethics, i.e. we as a humankind should ask ourselves a question: “Why Alternative Intelligence?” The answer to this question is an existential answer for the human, which intersects with our existence as a kind. Will Alternative Intelligence become the next necessary stage of development on the Earth or is it an optional branching? All of this should be considered in details, since the alternativeness of such formation as Alternative Intelligence prompts us that it is alternative to ours and it is yet unknown which dangers it might conceal.
- The possibility of existence of such an essence like Alternative Intelligence bearer within narrow limits of informational technologies is impossible. We need here a more complex approach, involving a wide range of sciences – biology, physiology, ……and certainly informational technologies. For example, the company Boston Dynamics demonstrates robot developments, which are able to realize mechanical functions of orientation in space for now. But the integration of organic researches, intersected with a social historical approach, i.e. integrating the principles of AI (Alternative Intelligence) may fundamentally change the production progress, generating social historical formations, and hence intelligence, but alternative to the human one.
We have considered only the most general consequences, which arise from the Alternative Intelligence, and it is necessary to take them into account when creating it. Of course there are even more of them, but the limits of one article do not make it possible to describe them.
Summary
While integrating History and Society into the problem of intelligence concept creation, receiving the concept of Alternative Intelligence we overcome the constraints and contradictions of Artificial Intelligence mentioned above and we receive a concrete scientific subject matter, which can be developed theoretically and practically.
This article is just a short essay and the main goal of it was to show the main contradictions of the approach to the intelligence creation and a proposal of a more adequate concept – the Alternative Intelligence.
References
- The Dialectics of the Abstract and the Concrete in theoretically-scientific thinking; Ilyenkov, 1997
- Dialectical Logic, Essays on its History and Theory; Ilyenkov, 1984
- Dialectical Logic; Orudzev,Sheptulin; 1986.
- Philosophy and culture; Ilyenkov , 1991
- Dialectics as theory of progression; Bosenko ; 1966
- «Artificial General Intelligence» Ben Goertzel, Cassio Pennachin (Eds.); Springer-Verlag Berlin Heidelberg 2007
- «Artificial Intelligence: A Modern Approach» Stuart Russell, Peter Norvig; Pearson, 3 edition (December 11, 2009)
- «The Philosophy of AI and the AI of Philosophy», John McCarthy; 25 June 2006
- The Origin of the Family, Private Property and the State; Engels, 1884