Artificial intelligence has recently been a hot topic. Mentioned in all configurations, used or planned to be used in almost every industry imaginable, it has also come to the film industry. It is not merely the subject of screenplays featuring fictional machines that try to take over the world, it is a part of the reality, a dynamically developing technology that can be used in film industry, or in media and entertainment in general.
Obviously, this technology has been in place in some aspects of the film industry for some time already. The most tedious, repetitive or time-consuming tasks, such as adjusting and filling parts of the background with colour or making corrections to film frames, are already performed using dedicated software. As a result, those tasks take much less time, and the effects achieved are at least as good as what is achieved by human beings.
In this year’s Avengers: Infinity War, the use of artificial intelligence was even more substantial. The face of Thanos reflected the facial expressions of the actor on an unprecedented scale. Effect of the work herein. This was achieved thanks to Masquarade, a custom piece of software based on machine learning algorithms. With Masquarade, the facial expressions of computer-generated avatar became much more natural and confusingly similar to the expressions of the actor.
Artificial intelligence and its potential business applications involve, however, various legal aspects which have to be taken into account. One of the most important issue is the question of authorship of the works generated by or with AI software. Resolving this question is important insofar as such works may be very successful and carry a measurable financial value. Such works are already being produced. Recently, for example, IBM Watson (the computer which in 2011 defeated humans in famous game show “Jeopardy!”) created the trailer for the film Morgan. Effect is incredible – see more here. The result of the computer’s work is phenomenal.
Since AI software is used in many more areas than merely analysis and preparation of “intermediate products” with a view to taking part in the creation of the final product, we have to consider what laws will be applicable to such software, and who will be deemed the author of the resulting work. All of this can come down to one fundamental (or rather philosophical) question: Can a piece of computer software or a machine itself be considered an author? Can it have its own rights in the works it creates? Accordingly, does the use of such works require copyright assignment or a license? If so, from whom?
To use a specific example, last year saw artificial intelligence being used for the purpose of writing a screenplay for It’s No Game – a film about AI. The film was directed by Oscar Sharp and based on a screenplay that had been written by a computer program that introduced himself as Benjamin. The program was created by Ross Goodwin, who also “taught” the program (by providing it with the adequate kind of data) how to develop screenplays based on traditional films.
Based on the copyright laws currently applicable in EU Member States, there is no doubt that only human beings can have copyrights. In line with this prevailing idea, the creation of a work entails a certain intellectual effort on the part of its author, and the work itself bears the author’s “personal stamp”. These factors contribute to its unique nature. The Polish Copyright and Related Rights Act specifies that it is the author who is the owner of copyright, unless the law provides otherwise. It is also presumed that the author is the person whose authorship has been announced to the public in any other manner in connection with communication of the work to the public. This, however, still does not resolve the issue of who in the It’s No Game example could be considered the author or joint author of the screenplay in question, as there is no doubt that the screenplay was written by a program. This program, however, could not exist without the person who developed it and who made decisions with respect to many factors affecting the final product. The person who fed the program with the right kind of data plays the same if not greater role here, since it is the analysis of these data that serves as the basis for how the program has an idea of the works it will later be developing. Thus, feeding the program with data on Western films vs. data on romantic comedies would yield two entirely different results.
So far, the only country which has a legal system that provides for computer-generated works is the United Kingdom. In this system, the author is “taken to be the person by whom the arrangements necessary for the creation of the work are undertaken”. This provision therefore stipulates that software and machines can “take part” in the process of creating a work. Nonetheless, here, too, only human beings can enjoy copyright. The provision in question was tested in the context of a UK case Nova Productions Ltd v Mazooma Games Ltd. In this case, the court held that the authorship of a work generated by a game engine (to be precise, the authorship of individual frames displayed to the player) has to be attributed to the software (game) developer. It should be noted that, under the facts of the case, the player was not guided by any desire to create such works so that only the designer of the underlying mechanism could be considered the author.
Despite numerous discussions and debates, the issue of computer-generated works continues to be governed by law that does not accept non-humans as authors. Yet, the questions asked are still pertinent, with no chance for us at producing a universal answer. The debate over whether authorship can be attributed to entities other than natural persons has been going on for almost 30 years, bringing no changes to the relevant laws. What is most often raised in favour of reserving authorship only for human beings are historical references to the origins of copyright and arguments emphasising creativity as a necessary element of authorship which stems from the human factor. However, the debate is growing more important now that we have seen the first court decisions on computer-generated works, like those mentioned above. Perhaps the status quo of authorship being reserved solely for humans will soon be challenged. For sure, the ever wider scope of AI application in creative fields will require revisions of many legal systems, including the EU system. Maybe the possible solution that would be adequate for the current state of the art in the range of artificial intelligence technology is leaving the authorship to human beings as it is established and providing part of the copyright of economic nature to the entity that is bearing the financial risk of such undertaking. The Legal Commission of the European Parliament has been debating this matter already for quite some time.
The result of the Commission’s work may have an impact on the film industry. In the near future, technologies might exert an even greater influence on the whole industry, giving rise to more and more problems. Technology will no longer be applied merely to reduce job completion time, increase accuracy or achieve a better final product. Already now, artificial intelligence governs many parts of the filmmaking process, such as predicting revenue (example here) analysing scripts or creating audience profiles for better ad targeting. Some people think that artificial intelligence will gain predominant influence on shaping screenplays (see more here) in order to maximise profits (if you want to check it – click here).
Check what we refer to:
- The Copyrights and Related Rights Act of 4 February 1994 (Journal of Laws of 2017, item 880, as amended);
- Judgement in case Nova Productions Ltd v Mazooma Games Ltd  EWHC 24 (Ch.)