Legal conundrums arising from works created by artificial intelligence
The major challenge that will be faced by AI works in getting protection may arise from conflict of laws between countries
By Anil Lale
Time and tide wait for no one and the same also holds true for Artificial Intelligence which has been growing at a breakneck speed. In the simplest of words, Artificial Intelligence a.k.a. “AI” is the capability of a machine to think like a human being. This is achieved by way of an artificial neural network which attempts to emulate the cognitive capabilities of the human brain. The evolution of humans as all of us know has been marked with the development of cognitive capabilities. This evolutionary process from Homo Erectus to Homo Sapiens (the highly intelligent primates) took millions of years and from Homo Sapiens to the modern human beings took thousands of years. Compared to that the evolution of machines from being tools for creators to be creators themselves has been at a very fast pace since the year 1956 when the term “artificial intelligence” was first coined.
We have been hearing about the awakening of machines from the days of the first Terminator movie in which the plot revolves around how an AI system takes over and divides the world in a war between machines and its human masters. However, AI has been the pet of science fiction than practical reality till a few years back. Since last eight to ten years, the growth in the field of AI has been on steroids. AI has transcended into our everyday lives whether in the form of a recommendation engine which suggests what next to watch on Netflix to self-driving cars. Big social media platforms like Twitter, Facebook, Instagram use AI in fighting inappropriate or racist content, learn user preferences, target advertising, fight spam etc. Nowadays, a chat with Siri or Alexa with its highly accurate natural language processing abilities will leave us wondering whether we are chatting with a real person or a machine.
The growth of AI has not only led way to a smarter way of life but also ideological collisions between smart people. When Elon Musk pressed the doomsday alarm by making a statement that “his exposure to AI technology suggests it poses a fundamental risk to the existence of human civilization” and making an appeal for proactive regulations of AI, Mark Zuckerberg responded saying that “Musk is a naysayer” and accused his doomsday fears of unnecessary negativity”. This debate between the Silicon Valley giants may have been consumed by us more like a tabloid capsule, but the basic conflicts which arises from this debate has several facets with various practical questions inscribed onto it. Presently, the growth in AI leading to technological singularity as cautioned by Elon Musk and Stephen Hawkins remains purely hypothetical and more of fiction. While the ethical, moral and existential questions arising from the development of AI and its cognitive capabilities are being debated, another facet which requires a more practical approach and immediate attention is regarding the literary, artistic, musical and other works created by an AI without the interference of a human. In such cases “Who will be the author of a work created by an AI?”
While computer generated artwork has been attempted since 1970’s, only in the recent past, the level of creativity demonstrated by AI has started nearing human creativity and examples of these are in front of us. In 2016 a short novel written by a Japanese AI reached the second round of a national literary prize and the AI created by the company Deep Mind has been composing music. Many will argue that such works created by AI lack originality required under Copyright Law as AI works are based on datasets which are fed to it. For example, if an AI has to create a musical composition, it first needs to listen to a lot of existing musical compositions and through reinforcement learning, the algorithms of the AI learns the patterns which it then uses to create a new musical composition. While AI may be mimicking the style of a particular composer by analysing the music created by such a composer, it is not recreating or using the musical compositions created by such composer in the past. Also, AI can analyse music created by various composers and develop a different combined style as well. What an AI is at present not capable of doing is to develop an original style which is not a combination of styles of different composers available as part of datasets to the AI. Would that mean that the musical compositions created by an AI is not original? There are various human composers who also emulate the style of famous composers and create music. If they can enjoy the fruits of Copyright, why can’t an AI created work also receive the same amount of protection when it demonstrates similar levels of creativity?
Generally under Copyright laws, only actual works (i.e. the literary, musical, artistic etc.) are protected and not an idea or style which underlies any of these works. So if the expression of the work is original (i.e. it is independently created without copying in full or part any pre-existing work) and it passes the test of “modicum of creativity” (i.e. minimal degree of creativity), the work qualifies for protection under the Copyright laws. To clarify further, to satisfy this requirement, the requisite level of creativity is extremely low and even a “slight amount” of creative expression will suffice. So more than the challenge of originality, what a work created by AI faces today is the challenge of authorship which leads to the challenge of ownership. Who owns a work that is not authored by a human and which is not the fruit of human intellect?
The U.S. Copyright Office has clarified that it will only register an original work of authorship, provided that the work was created by a human being. It goes on to state that copyright law only protects “the fruits of intellectual labor” that “are founded in the creative powers of the mind.” Because copyright law is limited to “original intellectual conceptions of the author,” the Office will refuse to register a claim if it determines that a human being did not create the work. A bit more digging into the compendium published by the U.S. Copyright Office will evince the fact that U.S. Copyright law expects machines only to operate randomly or automatically, if there is no creative input or intervention from a human author. The examples the U.S. Copyright Office cited in the compendium to cement this point of randomness or automation in a machine created work are “Reducing or enlarging the size of a pre-existing work of authorship, making changes to a pre-existing work of authorship that are dictated by manufacturing or materials requirements, converting a work from analog to digital format, such as transferring a motion picture from VHS to DVD” etc. So it appears that the U.S. Copyright law does not consider the possibility of an intelligent machine capable of creating an original work with modicum of creativity. That may also be the reason why a machine generated work is equated under the said Compendium to a photograph taken by a monkey or a mural painted by an elephant.
While the above is the approach under the US Copyright Law, Section 9(3) of the UK’s Copyright, Designs and Patents Act 1988 (CDPA) provides that the author of a work in the case of a work which is computer-generated, shall be taken to be the person by whom the arrangements necessary for the creation of the work are undertaken. Further Sec. 178 of the CDPA defines “computer-generated”, in relation to a work, as a work that is generated by computer in circumstances such that there is no human author of the work. These two definitions under the CDPA show the farsighted approach of the British legislature which back in 1988 had the vision to foresee that law may have to deal with circumstances where computers (i.e. machines) create work in such a manner that there would not be any human author. This approach answers both the questions above in affirmative and confers the authorship on the person (i.e. the human or corporate entity) on account of efforts and expense put in by the person for the purpose of creation of such work.
The approach taken by UK is also partially adopted under the Indian Copyright Act of 1957 where sub section (vi), which was added to Sec. 2(d) by way of an amendment through Act 38 of 1994, provides that the author in relation to any work which is computer-generated will be considered as the person who causes such work to be created. However, under the Copyright Act of India, the word “computer-generated” has not been defined like in the case of the CDPA. So the intent of the legislature is not clear as to whether it refers to works generated by a computer with the cognitive involvement of a human author or like in the case of CDPA it refers to the creation of work solely by a computer without human involvement. This lack of clarity is more striking because the word “computer programme” is defined under Sec. 2 (ffc) as a set of instructions expressed in words, codes, schemes or in any other form, including a machine readable medium, capable of causing a computer to perform a particular task or achieve a particular result. So this could lead to an interpretation that the law in India currently only deals with machines which follows a set of instructions to achieve a particular result and not machines capable of thought and creativity on its own.
The US, UK and Indian laws represent the three kinds of approaches which creations by AI may face globally. One clearly states that AI created work does not get the protection of Copyright while the other one clearly protects such works under its regime and the third one is ambiguous thereby potentially causing conflicting interpretations regarding copyright protection.
With these differences in approach taken by laws of different countries towards the same issue, the next questions which beg answers are “Whether an AI generated work needs to be given protection of Copyright?” and “What are the challenges in giving such Copyright protection?”
The answer to the first question lies in the very concept of Copyright itself. The need for Copyright protection first arose with the invention of the printing press which was the first means to copy a work in a mass scale. With this, the need to protect the exclusive commercial exploitation rights of the author of the work also arose so that such protection incentivises the creation of more works. So Copyright Law is a piece of legislation which exists in the first place to inspire creativity. We have in the past seen wonderful creations by great artists, authors, composers, lyricists etc. and now we are at the cusp of a new age where creativity can reach even higher levels with artificial intelligence competing with human intelligence. However, for this to happen and for other advancements in artificial intelligence, it is important for investments of time, money and sweat in this field. Such investment will happen only if a protection of commercial exploitation is available to the investor who causes the AI to create such works.
The major challenge that will be faced by AI works in getting protection may arise from a conflict of laws between countries. While some countries may give Copyright protection under its laws to AI created works, some other countries may decide not to. This conflict will prove to be a stumbling block for such creations as today content is created for the world and is exploited through satellites and internet platforms which is beyond jurisdictional boundaries. So if any country decides not to afford Copyright protection to works created by AI, then such works will turn into an open game for anyone else in that country to infringe and exploit. The answer to this will have to be the involvement of international organisations like WIPO and evolution of uniform laws across member states and through treaties with others. WIPO is presently leading the conversation on Artificial Intelligence and Intellectual Property Policy. While the scope of this is much larger and include inventions, some of the questions before WIPO are also similar to what we discussed above:
Does AI innovation and creation need IP incentives? Does the advent of AI require any changes to the existing IP frameworks? How should the value of human invention and creation be balanced against AI innovation and creation?
This is one conversation which is not just about Copyright but also about the first of many battles between human intellect and machine intellect. It remains to be seen whether human creativity wins over the brute force of the machine’s analytical and predictive capabilities or whether law be required to step into balance the equation.