AI Generated Output May Fail The Test Of Originality

AI GENERATED OUTPUT MAY FAIL THE TEST OF ORIGINALITY

Under the ‘minimal creativity’ regime, copyright protection is available if in addition to labour and capital, non-trivial skill and judgement is also applied

Copyright subsists in original works including literary, dramatic, musical and artistic works.¹ Here it is important to consider the meaning of the word, ‘original’ as otherwise, copyright protection will not be available.

The standard of originality as per requirements of the Indian Copyright Act, 1957 (hereafter as ‘the Act’) does not require the literary works to be inventive or novel. Instead, the Act requires that the literary works must be original in the sense that they are independent creations, and not copied from any other work.²

For e.g., someone may use an idea ‘ABCD’ and write a novel ‘12345’, and another author may use the same idea ‘ABCD’ and write a different novel ‘456781’. Two things can be said here. Firstly, the second author used a different expression from that of the first author. Secondly, the second author’s novel, ‘456781’ was written without copying the first author’s novel. Because of the two reasons, the second author’s work is considered as ‘original work’, even though it uses the same idea used by the first author. As the second author independently ‘originated’ the novel, ‘456781’ without copying the novel, ‘12345’, the Act will afford copyright protection.

However, the above definition of ‘originality’ ran into problems when confronted with the question of granting copyright protection to commonplace information. Even though some creator may independently curate and prepare databases and compilations, they may not be considered as ‘original works’, because such works are derived from pre-existing works that are already in the public domain, hence ‘commonplace’. They are also known as derivative works.

For e.g., consider a database labelled, ‘States of India’, in a simple tabulated format containing information regarding all states of India and their respective capitals arranged in alphabetical order. Suppose person A then rearranges the database without changing the layout or the information. Person Amerely rearranges the database along the north to south geographical placement of the states. Person A labels it as ‘States of India (North to South)’ Should person A then enjoy copyright protection? The answer is no because there is no creativity in mere re-arrangement of commonplace information.

For a long period of time, courts in India considered works as original to an author and allowed copyright protection even in cases of mere mediocre aggregation of data including publicly available facts, on the ground that compilation and aggregation involved tremendous investment of labour and time, even though skill and original judgement may be minimal or even absent.³ The idea underlying such a judicial position was known as the ‘sweat of the brow’ principle, as per which copyright protection was available because labour and capital was applied. This is a form of the labour theory argued by John Locke in the 17th century.

However, a shift has been visible, at least from 2008 onwards, when the Hon’ble SC held in the case of EBC v. D.B. Modak⁴, that there has to be some amount of ‘minimal creativity’ in order to qualify a work as capable of being protected under copyright law. Mere application of trivial or negligible skill, or application of merely labour and capital to pre-existing work will make the subsequent work only a ‘derivative’ of the pre-existing work.⁵

Under the ‘minimal creativity’ regime, copyright protection is available if in addition to labour and capital, non-trivial skill and judgement is also applied. This author submits that the element of creativity, i.e., effort required to produce a derivative work different in character from pre-existing works, is implicit in the application of judgment of the specific purpose of production. Application of non-trivial skill and judgement implies existence of a natural person who controls the final shape of the work. This is a form of the personality theory of copyright, as enriched by Hegelian ideas of works being an outcome of an author’s personality.⁶

The principle laid down in EBC’s case had been followed subsequently in the cases of Reckeweg v. Adven⁷, and University of Cambridge v B.D. Bhandari⁸. In Tech Plus Media (P) Ltd. v Jyoti Janda⁹, Navigators Logistics Ltd. v Kashif Qureshi¹⁰ and Satish Kumar v Khushboo Singh¹¹, customer databases were held by the Hon’ble Delhi HC to not satisfy the EBC’s original standards.¹²

When humans study or research or prepare a compilation of commonplace materials and produce works, they are capable of interpreting the data at hand in multiple ways, serving multiple purposes. The ultimate choice and autonomy regarding the interpretations to be adopted, the purposes to be met, and the arrangement of content remains with the human. It is the open-endedness involved in this endeavour, which makes the process a creative one.

Can the same standard of creativity be extended over to AI? To formulate an answer, it is important to evaluate the nature of AI and the way it functions.

AI is pre-trained and also trained how to re-train, in its business of finding, organizing and arranging source materials by combing through a pre-fixed database. AI is also trained on how to arrange the source materials that AI finds relevant and appropriate to the human querist. While the nature of AI’s output depends on the human querist, the labour and effort that the human querist would have to expend individually in the process of gathering data is saved by the AI.

This means that there is no scope for any application of creativity, because the judicial meaning of creativity’ requires application of non-trivial skill and judgement to the ‘arrangement’ of materials. Any ‘improvement’ that appears to be super-added by the AI is pre-defined as per the AI’s coding. While the AI succeeds in searching, selecting and arranging materials from a vast database, however, because of the pre-fixed way in which it functions, it cannot be said that the AI exercises any non-trivial skill because the AI produces output as per defined parameters and codes, which eliminate the scope of any ‘intellectual effort’ specific to the output. Nor can it be said that the AI exercises any judgment because of the absence of any ‘purpose’ specific to the output it produces. If it can be said that there exists a purpose specific to the output, at best, it bears an indirect link to the querist’s purpose, as the output is somewhat related to the query, and that too, not in all instances.

Disclaimer – The information provided above is general in scope and is not to be taken as a substitute for professional legal advise for any facts or circumstances specific to the reader. Discretion is advised