A Landscape of Artificial Intelligence in Canada
by C. Aubin & J. Freedin
Modern Artificial Intelligence (“AI”) is poised to take over human jobs in various sectors. Should artists and inventors be worried? How might Intellectual Property (“IP”) generated using AI be handled in the current Canadian legal framework?
AI is an umbrella term in computer science which encompasses many different computational techniques, including deep learning and neural networks. The common goal of these techniques is to simulate intelligence in order to solve complex problems, such as learning to perform tasks which were traditionally only within the grasp of humans.
Though historically a popular topic of science fiction, AI is a reality and has existed in one form or another for quite some time. Recent technological advancements have made AI more practical and accessible than ever, allowing it to find its way into some of our favorite technologies, including self-driving cars, personal digital assistants, and internet search engines. AI has also had an important impact in the medical sector, for example by aiding doctors in making faster and more accurate diagnoses.
The ever growing importance of AI has stimulated research in Canada and throughout the world. In particular, Montreal, home to internationally renowned experts in AI, has stood out as an important research hub in the field. The Canadian federal government has recognized the importance of AI in economic growth, and invested a considerable sum in Montreal’s AI research ecosystem. Montreal has also attracted interest from big players in the private sector. Google is building a new lab dedicated to deep learning in Montreal and has donated 4.5 million to the University of Montreal’s Institute for Learning Algorithms, and Microsoft recently acquired a Montreal deep learning research lab and invested in a prominent Montreal-based AI incubator.
IP in the Era of Artificial Intelligence
As in any emerging field, AI research is leading to the rapid generation of new IP with significant commercial value. Many large companies are actively expanding their IP portfolio in this field. For example, in recent years, tech giants such as Google, Microsoft, IBM, Amazon, and Facebook have increased their patent filings in AI-related technologies. These patent filings mainly relate to the technology behind AI systems, and are the fruit of R&D efforts by their scientists and engineers.
Unlike other fields, however, AI has the particularity in that the very use of the technology has the potential of creating IP of its own. In the Era of AI, IP is no longer generated by humans alone – it is also generated by intelligent machines driven by AI technology.
AI has been applied to create literary and artistic works, including using natural language processing techniques to create poetry, and using neural networks to create original (and sometimes unsettling) images. AI has also been applied to write computer code, and has even been successfully applied to design its own AI software which outperformed software designed by humans.
AI has also been applied to take on the role of an inventor. For example, an AI machine named “The Creativity” has been credited with creating new cross-bristle toothbrush designs, new ultra-strong materials, and devices that search the Internet for messages from terrorists. Genetic Programming, an AI technique which solves problems by evolving code using the principles of natural selection over many simulated generations, has been credited with several instances where it has independently created an entity replicating existing patented technology or has created an entirely new potentially patentable invention.
The Canadian Legal Framework
IP generated by AI raises important questions not yet specifically addressed in Canadian law. It is not unusual for humans to use computing tools to help with creating artistic works or designing an invention. However, as AI becomes more and more involved in the creative and inventive processes – up to a point where it could even be self-sufficient – one has to ask whether or not IP rights can even exist in the resulting creations.
Copyright: Skill and Judgment
Artistic and literary works are protected in Canada under the Copyright Act. This protection subsists in every “original” work, a qualifier not synonymous with “creativity” according to a 2004 decision from the Supreme Court of Canada. In this decision, the Court ruled that an original work must be the product of an author’s skill and judgement, meaning that the work must show “the use of one’s knowledge, developed aptitude or practised ability in producing the work”, as well as the “use of one’s capacity for discernment or ability to form an opinion or evaluation by comparing different possible options in producing the work”. This exercise necessarily involves intellectual effort and “must not be so trivial that it could be characterized as a purely mechanical exercise”.
Where a human’s knowledge and aptitude are required for AI software to generate a meaningful output (i.e. by providing inputs, setting appropriate operating parameters, etc.), the exercise may not be purely mechanical because of said human’s involvement, and a resulting work could be considered original. However, in cases where AI is autonomous and operates with little to no human input, the human’s contribution may be trivial, and it is uncertain as to whether the AI alone can be considered as having the necessary developed knowledge and capacity for discernment to create a work that would be considered “original”.
IP rights are intrinsically related to IP ownership. Indeed, without an author or an inventor, a work or an invention is not protectable. Under normal circumstances and in the absence of an assignment of rights or of a licence, copyrights and patent rights belong to the author(s) or inventor(s).
The use of AI may make it difficult to identify the actual author (or authors) of a work. On one hand, the user or the programmer of AI software may be considered an author if he or she has contributed sufficiently to the creation of the work. The input of users’ or programmers’ skill and judgement may thus raise questions of joint authorship.
In an alternative scenario where a work would be considered the result of the skill and judgment of an AI program as discussed above, the question also arises as to whether an AI program could be considered as an author under the Copyright Act. Such a question may be more philosophical than legal, as it requires determining whether or not intelligent machines can be deserving of the same rights as humans. An analogous issue was recently considered in the United States, where the Copyright Office specified that “a photograph taken by a monkey” is a non-copyrightable work after a federal judge concluded that U.S. copyright law “does not […] extend the concept of authorship or statutory standing to animals” and that “works created by animals are not entitled to copyright protection”. As animals are not recognized as having a legal personality under Canadian law for the moment, one could argue that an AI program could not be the owner of copyright following the reasoning of the US case. Following this logic, AI’s skill and judgement would not be sufficient to create an original work protectable by copyright as only a physical person can be an author.
Similar questions also arise in the context of inventions created using AI since an inventor must be a physical person under the under the Patent Act.
This is Just the Beginning…
The issues mentioned above merely scratch the surface of the potential challenges the Canadian legal system will face with respect to AI. We expect even more complex questions to arise as AI advances even further. Modern AI is considered “weak” or “narrow” in that it is only designed to solve specific problems, and cannot operate beyond what it has been programmed to do. However, we may soon achieve “strong” or “full” AI capable of performing a full range of human cognitive abilities, and perhaps being considered as potential equals to humans in the legal world and deserving ownership of IP rights on their own. While courts have not yet been seized with whether AI programs can independently generate or own IP rights, the question will inevitably come.