No final answer for the comparison question

Our focal starting point is the statement that the convergence problem between the human mind and machine intelligence is ill-posed. The statement is based on two lemmas:

1/ The problem is self-reflexive, a judgment where the judge, the human being is involved, s/he should decide on her/his state. The question is not helped by the fact that it refers to mankind as a whole, because mankind as a whole is questioned. That is a consequence of the Liar's paradox developed further in the Cantor-Russell problem. It was not science but art that tried to approach the above by satiric pictures and literature, e.g. Swift or the great Hungarian author, Karinthy.

A final and objective judgment can be expected only from a Being, outside and over mankind and human products. A minor remark: a mostly unconscious reason why mankind was always looking for such superhuman beings as final judges about truth and legitimation; that is the reason why I shift the final questions of ontology to the realm of theology. Ontology is understood in its original sense, meditating about the final problems of human life, the essence of human identity and the purpose of life. In computer science the word ontology is frequently used, or misused for categorization.

2. / The problem refers to an insurmountable amount of possible situations, reflections, and responses. We can quote a lot of cases where present machines are superior to the human mind and even more cases where convergence looks to be hopeless for the technology currently expectable. We possess no final answer concerning about technologies of the far future, nevertheless, any future technology success can provide with an answer related only to a final set of comparison cases. From logical point of view no perfect proof exists for all, infinite number of possible situations, questions. The attempt to state such a proof was started by Descartes in relation to man and animal and this was followed by the celebrated Turing test but neither of them and probably no further attempt could be successful. This is one consequence of the Goedel theorem.

Art tried - similarly to the first lemma - to approach that imagined situation-technology-response conundrum by science fiction. In this context we can refer only to expectable technologies and existing and accepted proving methods.

Discourse around vague concepts

The whole man-machine discourse is going around vague concepts and that makes any agreement impossible. No agreement exists on the definition of intelligence; it depends on progressive findings in brain research, ethology, social attitudes, prejudices, progress in computer science. The related dependence of any decision concerning human or man-man machine relations is also a self-reflective issue.

My working definition is, therefore, a relative one: intelligence is ability for responding to changing situations. In that mirror money machine is an intelligent device, amoebas are intelligent beings. A genius in mathematics can be minor to an illiterate child in some practicalities. The division between reason and instinct belongs to these uncertainties. In one aspect this is considered to be the separation line between animals and human beings; in another one, the same division is used for distinguishing between mediocre and talented people.

Similarly indefinite are the widely used concepts of consciousness and intentionality. They are based on subjective feelings and these feelings can be simulated to some extent by parallel working programs, one completing a certain task, the other forming a judgment about the task, the status and quality of the program and response to those qualia. To some extent, qualification should be emphasized, because a definitive comparison between a machine judgment and a human response is impossible, due to the above considerations. Even in our own activity can we hardly follow the separating lines of conscious and unconscious, intentional and reflex; retrospective analysis of our memory mixes those inevitably.

Here, we return to the same problem of the limited and the unlimited, similarly to the earlier considerations. If we define certain definite tasks for a theoretical metamachine, the comparison between machine and human behavior will be impossible because the machine can do what human beings "do". If we go over to the infinite multitude of possible human tasks, responses, the answer is impossible, due to the infinity of the tasks. I refer to the endless debate around Searle's Chinese Room metaphor and arguments of Penrose related to human instinct, ingenuity and machine reasoning. That is the reason why the old fantasies and argumentation about artificial and natural will never be completed. All those started with ancient mythologies, renewed in the Age of Reason, by Descartes and de la Mettrie and continued with the very start of computers not only by Turing or Neumann, but also by all participants of the hard and soft AI people.

Divergence is the real question

Until Modern Ages, i.e. until the advent of machines human activity was distributed by social strata. Most of the population had to do hard physical work and they had not too much time and energy for intellectual purposes. Even the activity of warriors was more bound to physical strength exertion rather than any kind of mind efforts. The change, as a result of machines, transferred all kinds of physical work to the new, non-human agents and the emerging computer technology does step by step the same with several mental activities. The first steps relate to tedious, routine tasks, e.g., copying, search, several simple financial operations, those executable by unambiguously definable logic programs. More and more sophisticated program systems do more and more work considered earlier to be bound to higher education and intelligence, higher than the average of the period. As examples, many kinds of design and diagnostic activities belong to that class. All boundary questions are heavily debated and are marked by changing views at each more or less relevant progress of technology. The most important discussion subject is the shift of responsibility, if it remains finally at the human operator or at the vendor or at the designer of the program? This example demonstrates the fundamental divergence of the problem: what will be delegated to automation and what remains with human beings? In other words: what will be machinoid and what will be considered as basically humanoid? Naturally, that is the dual of the convergence problem, but is much more important concerning our life, not the realm of science fiction and that of vague debates.

The example illustrates that the question relates to the basic problems of ethics, human culture, education systems, social relations, and individual self-realization.

Machines strip down all, or most in a certain context that is not human in a certain sense. Apparently, also these concepts are also weakly defined demonstrating the vagueness of the approaches. On the other hand, we can take for sure that everything that can be automated, will be transferred to machines and the activities that remain similar to the earlier physical efforts will have a significantly different role: leisure, like gardening, supply of mountain huts, handicraft.

What remains for the human being and what perspective is offered for future mankind? The pessimist scenario predicts the suicide of mankind. That suicide can be a virtual one: extinction of all human heritage, a degenerative intellectual and physical process, which can be the dramatic action of crazy criminals, fanatical groups. None of these kinds of scenarios can be excluded.

Neumann delivered a seminal talk about the subject half a century ago. The title was: Can we survive technology? He was an optimist in a realistic, skeptic sense. His point was historical experience according to which mankind survived several fundamental changes, some with great losses, torments but at last successfully adapted to the new situations. He recommended no panacea, he did not believe in any. On the other hand, he stated that "for progress there is no cure", i.e., nobody can stop the historical trend of development, in this case, the development of technology. With the passing of the last half-century all these predictions were strongly proved, also the failure of all wishful thinking attempts to stop or slow down the progress of technology. In our focus the combined technology of computers, information and automation plays the decisive role.

Accepting the conclusions of Neumann, we have to investigate what lies within our intervention capacity, how we can look at the ongoing and foreseeable changes as a process to be influenced in a moderate way. Before doing this let us have a look at those fundamental differences that are relevant from the point of view of divergence.

Two different phylogenies

Investigating the future of human-machine relations, we have to go back to very early preliminaries. Revolutionary results in biology prove the fact that all primary components of present life and human physiology are very early developments, several-hundred- million-year old ones. A similar finding demonstrates the history of our biological focus, the neural system and its central part, the brain. Kandel, who was awarded with Nobel Prize just recently, experimented with some mollusks and found in them all basic neural mechanisms of our brain, of course, in a rather primitive form. The meaning of the above for our context is that what we take as human is a complex heritage dating back to a terribly long time. This heritage carries all, or most of the remnants of earlier periods of evolution; it is built out of and upon them and is more or less activated in rather vague ways. Processing sensory inputs, sensitivities, emotions, vegetative and higher level reactions are all present and mixed with socialization imprints. The latter is also a collective product of the historical heritage, history means here human and prehuman history as well. Progress in ethology offers wonderful examples.

Our conceptual world is also a part of that evolutionary process and is based on the primitive sensory inputs and their early processing requirements, how to recognize similar or alien, friendly, useful or dangerous. These were important features in earlier environments that separated the basic concepts of aesthetics, logic and mathematics, abstraction evolved from practical learning conditions of survival.

On the one hand, the human mind could build an admirable world based on these primitives, a complexity that cannot be completely mastered either by itself (but can admire with some self-reference!). On the other hand, this infinite world is a limited one, only a subset of countless possibilities. The set of possibilities covers all other possible and not yet discovered worlds, whether those exist or not, and what is more important for us, all other possible interpretations, beyond the interpretation limits of our biological and mental construct. We do not know anything about these; even science fiction is limited to the frames of our conceptual world. We meet the difficulties in the perception of worlds different from our conceptual imprints, such as the world of quantum phenomena or of people socialized in an alien culture. In these encounters we try to apply a translation based on our conceptual frame but the perception is lame similarly to the translation of a poem in a very different language from a very different culture.

The working mechanisms of the neural system are still unclear in spite of very relevant findings that open many further problems. One thing is sure: the early electromechanical model of the McCulloch-Pitts relay-type is terribly far from the reality, one single neuron is a fantastic complex processing entity. If we try to apply models from the computer world, we could say that the brain works less like a logical machine of the general -computer scheme and more like a statistical Boltzman-machine.

An important remark: the application of computer metaphors has just the same limitations and the usefulness of the referred translation process: it covers the real phenomenon only by a limited and possibly imperfect representation but that is the way how we can approach the real phenomenon anyway; at best it can mirror some relevant features and lead to some useful further hypotheses.

The machine has a different phylogeny. It is a one time dedicated purpose device for solving some well-defined tasks, it is designed for adaptation to certain environments and to the execution of versatile objectives, it is limited in its working mode and activity area within a much narrower band than that of the human being. The performance of the components and the synthesis of the automata is well traceable and of nearly optimal design. None of these features are common to the physiological neural system. This is the real trade-off: a wide-band creature with a much less computable, reliable performance or a goal-oriented automaton more computable and more reliable.

Good and/or bad?

The question mark is most relevant. Ask a cooperative, self-reproductive automaton, a member of an imagined society of those products about good or bad. Suppose that it (?) can read and, in some representation, understand the history of mankind. The answer will be accompanied with terrible nausea (if it [?]) is created to have such).

Ask a devoted human observer. His summary: All the products of our cultural heritage will be lost if we drop the rather primordial emotions of aggresivity, social behavior towards the family, tribe, horde, nation, race, feelings regarding ''ours'' and ''alien'' sexual behavior of possession, race-preservation and all their further but not seceded sublimations.

Views on values are culture- and time-related. Even the value of human life has changed a lot. Capital punishment was accepted by all ethics based religions, social systems, now, fortunately, it is out of question in European civilization, not so much in the Europe- originated US civilization. Killing people of different religions, political views was a virtue also in the 20^th century. Nobody would now die for a king, what was the general rule not too long ago. Values related to sex have changed dramatically. Female virginity had a central value also in Europe. These and other values always played a decisive role in the selection and preservation of the parts of heritage that were enumerated in the divergence paragraphs.

The question is open in a wider sense than any time previously. Changes in civilization obliterated much of the earlier precious heritage and much of these revived, may be in changed forms, due to the primordial and historical-social imprinting of human nature, discussed under divergence. Nevertheless, our present civilization is the first to observe the whole past with a possibly detached view from the present values and images, in a possible but limited objective way. That change opens more freedom for selection. The other reason of the openness is somehow rooted in the freedom given by the machine world. The world of major physical efforts was a world of social compulsion due to the compulsion of survival-production demands. Though far from the ideal the freedom of people due to progressing automation is developing fast. Much more free time, much more freedom in selection of products, leisure, cultural opportunities, traveling are the landmarks. Good or bad, and in what directions, are open in many ways: open in the sense of value estimation, open in the sense of liberty concerning present individual desires, cultural backgrounds. Finally, the question is open about the possible change in the weight of different human driving forces and the possibilities of influencing that change.

Returning to the starting question: suicide, better mankind and better in what sense?

Present status

It is contradictory as everything and always. On the one hand, the progress of science reached never-expectable frontiers and opened vistas beyond those. Enumeration would be a commonplace. Mentioning the revolutionary change in the worldview might be more interesting a phenomenon referred to earlier. It is related to the relativity of social-cultural values. Relativity means a situation, circumstance, environmental dependency on phenomena having been viewed absolutely fixed earlier. Changes in worldviews took place several times in history but each one was certain in the final uniqueness of the achieved novel Weltanschauung. In spite of the terribly aggressive political ideologies of the 20th century, science rendered a totally new permissive look, a small hole on the closed frame of human conceptual world, mentioned above. Not by chance were these vistas and are still the most obscure realms for the everyday man, for those who are not deeply educated in the sciences and even for everyday scientists themselves. The unending disputes around interpretations are ongoing between the hard AI and those who believe in the uniqueness of the human mind, and that is only a minor part of that quarrel.

Our main lesson is the emergence of that relativity, pointing not towards new obscurantisms but towards a new, relativity-riched rationality. The dangers of new obscurantisms are present; in several forms they are devastating. The new rationality is a relevant defense against those. This new rationality applies the earlier and current results of mathematics and logic. In mathematics different approaches of uncertainty, theories about different descriptions of spaces, their transformations, i.e., modern algebra, topology are the instruments of the new, freer look. In logic the development of situation-dependent reasoning, modal and intentional logic, the new achievements in the relations of natural and formal languages have similar roles.

The confusion is rooted in the transient processes from a fixed-frame world to a relativistic one where the frame and the phenomenon are the interactive ingredients of the same infinity. That infinity is an open world but the new rationalism with its permissive pragmatism clears the way for establishing a ground to establish a relatively good environment for the present and maybe for the foreseeable short future, not longer than a human generation. The relatively good is now opposed to the absolute bad and that is a mark of change, too.

The confusion is best represented in the arts. Modern art representing the relativity of values got under way nearly simultaneously with modern science. That was an arena of many ephemeral happenings and of swindlers, too, but the majority reflected the deep responsibility and bewilderment feeling the uncertainty after the collapse of the imagined fixed frames. That bewilderment opened new opportunities and artistic methods to look inside, into the human psyche, a dual to the new vista of outside, by science.

Philosophy

Being an outsider, it can be a bold and impertinent venture to say anything about new roles of philosophy. What follow is a conclusion of the above considerations and a view of somebody who tries to organize the consequences of his look and experience in a problem area concerning his own research territory. That is the fate of most scientists after a period of narrower, more factual activity. Senior wisdom and senile silliness are present, mixed.

Philosophy has been suffering crisis for a long time, too. After the advancement of sciences less and less territory is given to those kinds of speculative word views, the integration that were the usual role of philosophy. Keeping the infinity of facts and unsolved problems in mind, the earlier integration has lost its credit; on the other hand, the natural drive for an integrated view, totality of explanations did not decrease; what is more, the drive captured most of the great pioneers of science.

Under the chapter about final questions in lemma 1. / I stated my view about the impossibility of a scientific ontology and my conviction that all major issues of ontology belong to the realm of theology, belief either somebody is a believer or not. Both standpoints should contain empathy and tolerance towards other beliefs up to the limit of the other's aggressivity.

The problem of epistemology, the other traditional branch of philosophy is more complicated. I tried to discuss the epistemic reflections on my field, computer science earlier. This led to the new conclusion about the relativity of epistemology. The statement is somehow a contradictio in adjecto, perception and representation of facts of nature should follow the same rules for truth, the same conditions of prove and refutation. That was for a long time a criterion of science as a totality of knowledge about the world and was strengthened by the great efforts of positivism. The most important ambition was the mathematical-logical formulation, the effort of Russell and Hilbert, not to mention the great antecedents back to Greek Antiquity and followers in philosophy of science, e.g., Popper. The backlash came with Kuhnian and further with the Feyerabendian overall relativity that is not identical with the rationalist relativity of this paper.

The science of a certain age can well reflect the part of the true phenomena that can be iteratively reproduced, practically applied, logically included into the frame of the measuring and processing techniques of the age. Here stands relativity, on the one hand, and a rather pragmatic but rigorously definable measure, on the other. If this rigor did not exist, no standards could be set for the application of any technology, medical practice, and social decision. Those advocating only the relativity side of this changed epistemic view, and are consequent enough, should throw away all achievements of science and go back to a state of mankind before any civilization. It is clear that the rationalist relativity standpoint is open for any novelty that can stand the above criteria, finally of usefulness in any practical or further aspect, on the other hand, is a defense against charlatanism and the antiscience trends.

The relativity-view, too, is a trap. Each branch of science has different standards of proof. Compare the epistemic standards of mathematics with those in history or sociology. In physics the epistemologies of cosmology, particle physics and turbulence are rather different. In experimental sciences that epistemic relativity is rooted in different reproducibility, sampling possibility, tolerance in clustering the identical and in accuracy of measuring methods. This latter, too, is a complex concept; accuracy from different points of view is mostly an intrinsically bounded feature of measuring methods, not only the final (?) Heisenbergian uncertainty.

The lesson is hard. All branches of science have somehow different epistemic views, not only methods. The views are, naturally, mostly derivatives of the methods. A general epistemology, covering all sciences, including humanities is one of the open questions, and one of the practically very important issues. On the other hand, we cannot be sure that such kind of generalization is possible at all. For the time being the epistemology of the individual sciences should be improved, and this is, first of all, not the task of philosophers but of the specialists of the field concerned. Efforts in the direction of generalization are real philosophical problems but in a much more humble way than some present philosophers think.

Aesthetics and ethics were mentioned in some respects. We arrive at the initial problem: more humanoid humans after the creation of more and more supporting machinoid machines. The human being, elevated and threatened simultaneously, is in the renewed focus. Several sciences work on the subject, e.g., neurology, other branches of physiology, psychology, linguistics, pedagogy, sociology, history and all kinds of interdisciplinary fields. Nevertheless, the human being is more internally bounded complex system rather than any other is. Maybe that is true only for us, but as a Hungarian poet, Babits says: "Compelled to be the hero of my verse, The first and last in every song I write".

Anthropology in a holistic sense can be the answer and the focus of our philosophical investigations. Not a new idea, most of the ancient Greek philosophy focused consciously on the same. The new view offers new challenges and the idea of more humanoid human supported by his/her creatures suggest a new need. Any integration is a hard task. The different disciplines must create a common language to communicate their results to the others. This language is very weak for the time being; even conceptual definitions are different, sometimes contradictory. The subject is really multidimensional; research people in different disciplines have a rather low-dimensional view, like our low-dimensional space imagination. A lot of humble, open, empathic dialogs are needed.

The author expresses his gratitude to Márta Fehér, Professor of Philosophy, for her careful revision of the text and relevant advice.

January 25, 2001