This post has been authored by Unmekh Padmabhushan, a final year student of National Law University, Jodhpur.
Machine learning is the process by which a piece of software is able to expand upon its capabilities and knowledge in a self-driven manner without any significant human input. This technology has been used, for example, in disaster warning systems and in driverless cars. Another scholarly use of such technology allows robots to derive patterns and significant correlations from enormous databases of texts in a manner impossible for human beings. This has led to led to an explosion in the ability of those working in the field of the humanities to analyse data like their natural sciences counterparts have done for years. 
Expression and copyrights
Copyright law limits its protection only to expressive works – ie. the expression of an idea rather than the idea itself. The Indian Copyright Act, 1957 defines neither idea nor expression and is also silent on any difference in the treatment of the two. However, in R.G. Anand v. Deluxe Films, the Supreme Court stated that there can be no copyright in an idea itself and that violation of the copyright is confined to the form, manner and arrangement and expression of the idea by the author of the copyright work.
Reproduction of a work ordinarily involves the purpose of engaging with the work as an end in itself – for example, reading a book or watching a film. However, a computer lacks this appreciation for and, therefore, the ability to enjoy the work in its true sense. Consequently, this is no threat to the substantial rights of the copyright holder and does not affect the normal exploitation of their copyrights.
Legal position on non-expressive use
A number of judicial decisions have recognised that such non-expressive use falls within fair use exceptions. In Sega v Accolade, using the entirety of console code for the functional purpose of reverse engineering compatible games was held to be fair use by the US 9th circuit court. Similarly, in Perfect10 v. Amazon., the fact that Google’s search engine did not make any artistic use of the stored images, but merely used them for reference became a critical reason for the eventual finding against a copyright infringement. Perhaps the judicial decision most relevant to the present issue came in Authors Guild v. Google Inc, the Google books case. The court stressed on the transformative effect of the manner in which google books contextualises its material. The court states that the program “may well harm authors’ markets, but such harms will generally occur in relation to interests that are not protected by the copyright.”
Section 52 of the Copyright Act, 1957 includes exceptions akin to fair use. The enumerated purposes under this section are restrictively interpreted because any use must necessarily fall within the scope of the defined uses. Therefore, as per §52 of the Copyright Act, the work must be for private study, research, criticism or review. Commercial use, such as for the purpose of improving translation services or for improving the pronunciation of assistants for the disabled would not fall within this exception.
Unlike the flexible, future oriented law in, which seems considerably better suited to make the best use of emerging technology, Indian law appears to be stuck in an inflexible straightjacket. In such a scenario, it would appear necessary to at least incorporate a specific provision that deals with ‘non-expressive’ fair use, in the nature of an exception to copyright infringement. The UK, which has a similar, close ended approach to free use as India, has adopted such a recourse through the insertion of §29A of the Copyright, Designs and Patents Act.
The submissions made to the Australian Law Commission on the issue highlight that the uncertainty surrounding caching and indexing inhibits innovation and investment in cloud computing and other products and services which rely on such incidental or technical uses. Data Mining and other such non-expressive uses include critical applications like Healthcare, Agriculture, and Banking. It is clear that such use can have a profound impact on the quality of life in a country and be the measure of its economy.
Therefore, India should abandon the dinosaur
of the closed system inherited from the English. The Australian
Law Reform Commission has recognized the importance of moving away from a
similarly archaic structure – in its report on incidental and technical uses,
it advocates for a fair use recognition of such non-expressive use, in the
manner of the American courts. Singapore too has shifted to such a method by
introducing an open-ended fair dealing exception under §35(2), Copyright Act 1987 in the year 2004. A flexible, principle based determination of fair use, rather than
a rigid formula is better suited to a rapidly evolving landscape for copyright
law brought on by technological changes, especially in cloud computing and AI.
Therefore, if India wishes to leverage its
technical know-how, and allow artists, academicians and companies to leverage
the complete power of artificial intelligence and text-mining applications, it
is necessary that an additional exception is added to Section 52 of the
Copyright Act, 1957. However, this can only serve as a temporary solution. As
the instant example shows, a close-ended fair use exception is simply too
laborious and cumbersome for the modern digital age, where rapidly changing
technology demands a broad, principle based fair use doctrine, which is capable
of being applied even to applications which were not envisaged at the time when
it was drafted.
 Matthew Jockers, Macroanalysis: Digital Methods for Literary History 3,4 (2013).
 Narayanan P, Copyright and Industrial Designs 8 (2002).