A.Yu. Alekseev - Ph.D. (Philos.), Leading Research Scientist, Philosophy Faculty, Lomonosov Moscow State University, Associate Professor, Moscow State Institute of Culture; Science Coordinator of research programs of the Scientific Council of Artificial Intelligence Methodology of RAS. E-mail: aa65@list.ru T. Pozarev - Post-graduate Student, Philosophy Faculty, Department of Philosophy and Metodology of Science, Lomonosov Moscow State University. E-mail: todor_pozarev@icloud.com

Lovelace test is supposed to answers the question \\\"whether the computer can create?\\\" and it discusses the architecture of "creative" computer. The Lovelace test is analysed in a number of works in which \\\"creativity\\\" is identified with the \\\"art\\\" ? lite-rature, poetry, music, paintings [2]. Into account are taken supposedly unlimited possibilities including Internet argument [3]. It is suggested axiomatic of the computer creativity [2] which essentially reproduces the Aristotelian idea that the creator is not able to explain and reproduce the creation of the artefact. The conclusion is that non computer is able to pass the Lovelace test, including "oracle machine" since it works in predefined context of it-s creator. In our work we support such argument but it stands only for symbolic computing. In noted works are ignored the possibility of neurocomputers. It is suggested neurocomputer version of Lovelace test which is followed by three paradoxes. Lovelace paradox (who is the author?). The paradox quite clearly formulated by Ada Lovelace in 1842. According to Lady Lovelace the computer is capable of producing artistic works for what is necessary that the computer executes signature of the algebra, music theory for example. However, the computer is not able to create new forms on it-s own, because the true creator is he programmer and machine is able to perform only predefined functions by designer (see [1]). Alan Turing paradox (how to imitate creativity?). Indirectly formulated in 1950 by Alan Turing, when refuting the argument of Lovelace as the main objection of computer conception of artificial intelligence. Paradox consists in the fact that "intellect" is imitated by computer. However, the \\\"creativity\\\" as a constitutive feature of intelligence denies any form of imitation because it is impossible to imitate "original". The Lovelace paradox is seemingly solvable if interactive process of auto-learning is imitated. But in that case it is necessary to program casual connections of auto-learning and imitate brain processes in of the child-machine as Turing suggested. Universal digital computer does not implement such a program. It is able to program a theory of formal neural networks, but that theory is representable algebraically, a priori known by designer and therefore it is not capable to produce nothing original and unique. Turing neural network cannot be reduced to this formalism, but nevertheless this has newer been tried. Turing had succeeded only to offer a number of phenomenological training rules [1, p.384]: designer form Lovelace test posses the general principles of creating computer artefact, and it is able to predict and guide the behaviour of the - child\\\". But designer in this case is not able to set specific parameters of this "creation" because of the complexity, continuity and dynamics of causal relationships. For the uniqueness and importance of the product is responsible other interactive component of the neurocomputer test - it-s "student". Putnam\\\'s paradox (how to express the inexpressible?). It stands that even in case that neural network of child-machine is created, is it possible universal self-learning program, which would implement general theory of creativity - Computer functionalism of Hilary Putnam (1960, 1967). states that mental abilities including creativity are produced by natural neurocomputers of animals and humans, by artificial neurocomputers of robots and imaginary "neurocomputers of Martians". In clean connectionist machine (ideal which represent Korsakov machine) functional specifications of universal scheme of that kind is not specified the language of an external designer. On the contrary, such language is formed from a functionalist specifications of dynamic link features, responsible for the specific conditions of self-training. Hilary Putnam considered in 1970 that the paradox of universal specification of unique phenomena is unsolvable since computer is not able to imitate biological and social phenomenas due to the complexity of causal connections and their ontological origins. However, within the framework of the Turing-s imitation of the creative process such a global task does not exist. \\\"Universal language\\\" in any concept of modern functionalism (psycho-functionalism, role-based and analytical functionalism) represent neutral language. However, between these two \\\"measurements\\\" is given only a priori connection. Interactive, computer-engineered con-nections are reviewed in three dimensional languages, based on a combination of machines Korsakov and Turing [4]. The Korsakov machine simulates inter level connection attributes that can be represented semantically dissimilar languages. Language connectionist specifications of creative phenomena are weakly correlated with both languages of humanities and languages of sciences, as well with natural language. In perspective, neurocomputing version of Lovelace test leads to one 3D-representation of descriptions: 1) neurocomputing base, 2) it-s creative consequence 3) to the connectionist program related tokens. Conclusion. Noted paradoxes of computer art - authorship, simulation and expressibility - according to our opinion are solvable in a self-learning connectionist machine. This work was financially supported by RHSF in grant № 15-03-00519а «Post-non-classic paradigm of artificial intellect».

 

1. Bringsjord S., Bello P., Ferrucci D. Creativity, the Turing Test, and the (Better) Lovelace Test. Minds Mach. 2001. February. V. 11.№ 1. P. 3-27. DOI=http://dx.doi.org/10.1023/A:1011206622741.
2. Mark O. Riedl. The Lovelac e 2.0 Test of Artificial Intelligence and Creativity / Proceedings of the AAAI Workshop: Beyond the Turing Test. Austin, Texas. 2015.
3. Alekseev A. JU., JAnkovskaja E. A. Fenomenologija kompjuternogo tvorchestva: test Ady Lavlejjs // Filosofija tvorchestva: materialy Vserossijjskojj nauchnojj konferencii, 8-9 aprelja 2015 g., Institut filosofii RAN, Moskva / Pod red. N.M. Smirnovojj, A.JU. Alekseeva. M.: IIntell. 2015. S. 376-394.
4. Alekseev A.JU.ZD-semantika slovarja kognitivnojj biomediciny i mashina Korsakova-Tjuringa // Biomedicinskaja radioehlektronika. 2015. № 4. S. 9-11.