AI CARAVAN MOVES ON. DOES IT NEED COPYRIGHT INCENTIVES?

Problem statement. Artificial intelligence (AI) handles everything, whether fueling extensive discussions or continuing developing at an exponential pace. Some people believe that we are witnessing the Fourth Industrial Revo lution, while others i) seek ways to curtail the growth rate of AI before it is too late or ii) simply acknowledge that it is already too late – Pandora’s box has been opened. AI industry has no time for this and pays little attention, it goes on. One wonders if augmentation of AI is dependent upon stimuli at all. If not, does it mean that artificial intelligence may become victim of its own success? Analysis of recent studies and publications. The copyrightability of AI-assisted and AI-generated objects is an issue that has been raised in many papers by various scholars, namely J. Cabay, P. Devarapalli, D.J. Gervais, J.C. Ginsburg, T. He, G. Huson, A. Michel, V.M. Palace, S. Ricketson, P. Samuelson, J. Wagner, J.M.N. Zatarain and many others. The agendas of different governmental and non-governmental organisations also include the question of AI and copyright, and IP in general. Nevertheless, the issues of protection and regime, suitable for the deve lopment of AI industry, remain open. The Article aims to ascertain whether there is a genuine and well-grounded demand for granting copyright incentives with respect to AI-results. Main text. Copyrightability of AI-assisted and AI-generated objects

К искусству нет готового пути. Будь небосвод и море только сини, -Ты мог бы небо с морем в магазине, Где краски продают, приобрести. 1 Samuil Marshak Copyright law is anthropocentric. 2 It mainly comes in originality and human author requirements. The international copyright law provides us with implicit evidences supporting this statement. The historical background of the Berne Convention for the Protection of Literary and Artistic Works and its interpretation support the idea that these requirements underlie its provisions. According to Professor Ricketson the human-centered notion underpins the following provisions of the Convention.
The Convention does not define the terms "author" and "intellectual creation" or "originality". Professor Ricketson considers that "there was nonetheless a basic agreement between the contracting states as to the meaning of the term (author), and, because of this, it was thought unnecessary to define it." 3 This is also true of intellectual creation. The proposals to enshrine this requirement in the Convention have been rejected on the ground that such a requirement is already inherent in the term "literary and artistic works." 4 Collections of literary or artistic works such as encyclopedias and anthologies are the only works for which the Convention establishes intellectual creation requirement to qualify for protection under copyright law (Art. 2(5) of the Convention). This is due to the fact that such objects can be created with or without manifestation of creativity. Moving forward, the appeal to the rights of authors means the appeal to personal, rather than corporate, rights. The term of protection is made dependent upon the life of the author (evidently, a human being). The introduction of moral rights was also justified by human needs. Finally, cinematographic works are the only objects which can be created by non-human authors. 5 Despite being implicit at the international level, these requirements are internationally applicable standards. At the national level they are usually explicitly stated in legislation and/or largely interpreted in court practice. 6 However, there may be considerable differences between the approaches as to the threshold of originality.
Bearing this in mind, AI-generated objects are subject to entering the public domain. AI-assisted objects present a different category. They are protected by copyright so long as artificial intelligence remains a tool provided that such objects meet other requirements. The transition of AI "to the next level" will make the objects also uncopyrightable.

AI growth: dependence upon v. independence of copyright incentives
Debates over AI-assisted and AI-generated objects often include the issue of AI reliance on copyright incentives on the agenda. Is it true that AI further development is dependent upon the decision on granting copyright incentives for the respective objects? Considering that supply is coupled with demand, ascertaining whether there is genuine demand for such incentives should be the starting point.
The proponents of so-called dependence theory would suggest that every creation should count, whether created by human or not, especially when it is not discernable who is behind it. Another argument lies within marketplace competition: free AI-results may distort competition; affording copyright protection to them would level the commercial playing field and create conditions for competitiveness of human results. 7 Even if the proponents of the above theory give ground on their initial position, they will keep insisting that the industry, particularly its larger system of production, is driven by the pursuit of financial gain 8 and depends upon stimuli. Introduction of work-for-hire doctrine in the US serves as a sign that copyright can make some kind of concession -the door opens to one who knocks. AI industry has got the hint clearly and paves the way. Champions of AI development claim that if the door remains closed and all AI outputs enter into public domain, this will disincentivize others from reinvesting in AI research. 9 The proponents of so-called independence theory would refute the above statements. They insist that AI needs no copyright incentives to produce results. 10 It is one thing to motivate and inspire people to create works through economic and moral rights. AI does not wait for inspiration to strike. At this juncture we pause to paraphrase a famous proverb and say that AI is as AI does. In this Article when we mention AI, in most cases we mean "weak" or "narrow" AI which functions in a narrow area and remains confined to it, though possessing autonomy and demonstrating results sometimes outperforming human ones. The current level of AI allows us to say that AI does not even understand the meaning of the output it is creating 11 because it still lacks conscience that is the attribute required for such activity. All it takes is electricity (or some other motive force) to get the technologies such as AI into production. 12 Therefore, AI could, in theory, never stop creating regardless of the AI not receiving compensation for its work 13 or any other economic encouragement, or not gaining reputation points. 14 Addressing the issue of copyright incentives for AI will force us, among other things, to make a distinction between "production" and "creation". Following Kant's moral proof, creativity goes beyond logic. Art may develop but it is not path from simple to complex -art explores new styles and movements through both widening and deepening. The fact that AI can produce facially originally objects does not mean that the production process is creative. In this regard, the question of method, the subjective or objective standard, 15 and the interpretation of originality become extremely relevant. Before embarking on any side, we should notice that human-authored pieces and AI-generated content are different. 16 People are intrinsically creative; they are powered by intrinsic motives and desires. 17 As regards AI, making new objects falls within its functioning and does not result from its own desire or intention. AI is bounded by algorithms, features and patterns; its outputs, which are subject to the influence of environment, are still determined by the inputs leaving no room for creativity. This refers us again to the originality requirement.
Having established that AI produces rather than creates, we have to admit that the scale and capacities of such production impress and even raise concerns. Using AI, it is already possible to create all possible variants and combinations within certain parameters. 18 This means that AI can exhaust all expressions of the idea. Affording copyright protection to AI-results would destroy the fair balance. 19 Moreover, rights come with duties and responsibilities. The same applies to copyright. AI cannot gain copyright rights unless it, as purported "author" (as a matter of copyright law), can accept full responsibility for "its" creation, meaning and content. If suggested as an alternative, appointing a human proxy "author" will also run up against a problem, namely the one of identifying the person who deserves such status. 20 We may return to issue of AI as an author, if we encounter "strong" or "general" AI. Should this happen, call for incentives will pale in AI personhood deliberations. Unless law wants to be mired in disputes, it should take a proactive role now and elaborate a solution, satisfying current needs as well as future ones.
While discussions are in progress, artificial intelligence, though in the absence of copyright incentives, shows no signs of stopping as well as is likely to continue flourishing which is probably due to the incentives inherent to the artificial intelligence industry. 21 Firstly, it is in the interest of business to continue running the race. Secondly, it is in the interest of states to fuel the race and participate in it.
We may deduce that the technology itself does not need copyright incentives. Does the same conclusion apply to stakeholders of AI industry, particularly AI programmers and investors? The proponents of so-called independence theory would answer this question in the affirmative. Investors provide no creative contribution necessary to consider them among "applicants" for copyright incentives. As for AI programmers, firstly, given the machine learning process, the results assisted or generated by AI are at least one degree removed from the human programmer(s) of the AI code. 22 Secondly, the relationship between AI programmer and AI itself resembles the one between parent and child 23 or master and pupil. 24 AI programmers give rise to and determine the scope of AI, amounting to original code, features and input data used for training, but have nothing to do with the output.
Conclusions. If we face the truth, it is highly unlikely that absence of copyright incentives has significant negative impact on AI industry. With or without them, AI continues developing. Granting AI copyright incentives would be unjustified. It all comes down to the issue of nature: nature of copyright with its anthropocentric soul, nature of AI and nature of stakeholders' interests. In case of weak AI, which we are already witnessing and experi-encing in everyday life, requests for copyright incentives seem premature and artificial. They emanate from stakeholders, among which éminences grises are companies and corporations investing a great deal into the AI industry. However, the rationale behind their requests has nothing to do with creativity. It is therefore a different type of claim involving protection of investments.
Considering that the protection is claimed for investments and drawing a parallel with databases, sui generis protection appears to be the most attractive option. Its main advantage is the possibility to determine the content of the regime according to the existing needs, AI functional peculiarities and real and potential capacities.
Given that the discussion on AI leaves copyright dimension, such concessions as absence of moral rights and shortening terms of protection will not suffice. The sui generis AI right is expected to maintain the interest in developing the AI industry. However, it is even more important to disincentivize unfair practices. In case of AI-assisted or AI-generated objects, created as a result and/or for the purposes of abusing the right, such objects should receive no protection and be expressly excluded from sui generis right and any protection.