The Rise of the Machines

 

“The rise of the machines is here, but they do not come as conquerors, they come as creators.”[1]  This concept of creating is fundamental to Canadian copyright law.  Indeed, copyright’s raison d’etre has been described as “encouraging the dissemination of works…and obtaining a just reward for the creator.”[2]  Why then, have these machines divided many over the appropriate application of copyright to AI-generated works?  The problem lies in determining who is in fact the creator – or more aptly in the copyright context – the “author”.

 

This is because only original works may be the subject of copyright.  Although the threshold for originality is low (the requisite level of skill and judgement necessary to produce an original work need only exceed what might be characterized as “a purely mechanical exercise”[3]), it is generally accepted in Canada that “a human author is required to create an original work.”[4]  Recently, a U.S. district court decision reaffirmed this principle south of the border.[5]  The question then becomes, when considering the use of AI, where does human authorship end and non-human authorship begin?  This question has plagued me since Professor Festinger posed it on the first day of class.

 

On the one hand are those who believe that equating machines to creators conflates the means with the end.  From their perspective, “AI generators are nothing more than a modernized version of the tools artists have used for centuries.”[6]  This argument is compelling.  Surely, the use of technological aids is not the line demarcating human and non-human authorship.  Copyright is a product of technological innovation – and since the Statute of Anne, copyright law has absorbed most new technologies with ease, from the camera to the computer.  The question has never been whether a human author used a computer to produce the work, but rather, whether the computer “took over all the necessary decisions involving skill and judgment.”[7]

 

This was a central issue in Geophysical Services, where the Alberta Court of Queens Bench considered whether the use of computers to produce seismic data satisfied the human authorship requirement.  The court rejected the argument that the data was collected “by simply pushing record on the computer”.[8]  Instead, the court pointed to the fact that it was humans who were directing the computers and other technology.  It was found that the crew used “expert scientific skill and judgement”.[9]  Skill was involved in planning where to acquire the data and knowing how to use the technology.  Judgement arose where the seismic crew compared possible options and checked the data for its quality – “to the extent that there [were] problems…the process may have [been] restarted with different parameters.”[10]  Therefore, although the information was in fact generated by the technological tools, there was sufficient human intervention to find human authorship.

 

Similar arguments could be made for the use of AI in generating works.  As far as I understand, AI-generators do not create works on their own.  A human directs the machine.  In fact, AI prompt engineering has been recognized by some as “the most important job skill of the century.”[11]  Indeed, it is a skill already highly sought after by employers across a vast array of industries.[12]  Thus, it appears that human skill is necessary in planning how to prompt the AI and knowing how to use the technology.  ChatGPT says the following: “crafting a well-formed and specific prompt can significantly impact the quality and relevance of the AI’s response… prompting AI effectively is not just about technical knowledge but also involves an understanding of language nuances, context, and the specific capabilities of the AI model. Practicing and refining this skill can lead to more productive and accurate interactions with AI systems.”[13]

 

Thus, prompting, re-prompting and reviewing AI-generated works could very well amount to sufficient human intervention under the reasoning in Geophysical Services.  The algorithmic processes in response to human directives are merely mechanical exercises.  It is the use of the tool, however, that requires human skill and judgement.  This argument is bolstered by the principle of technological neutrality, which holds that copyright law should not be interpreted to favour or discriminate against any particular form of technology – “it protects authors and users by ensuring that works attract the same rights…regardless of the technological means used.”[14]

 

Although I find merit in this perspective, it does not sit comfortably.  After reading an article written by Carys Craig[15], I began to understand why – it fails to see the forest for the trees.  She observes how the AI era provides an opportunity to revisit the purposes of copyright: to incentivize authorship and the dissemination of original works.  These objectives must inform any application of the principle of technological neutrality, for it is the copyright itself that must remain technologically neutral.  Simply treating things equally, regardless of there severe differences, may produce undesirable results.  Armed with this substantive approach to technological neutrality, Craig proposes that granting AI-generated works copyright protection would neither reward the people responsible for the AI, nor would it encourage the authorial creativity and progress that lies at the heart of copyright.  To do so would “cause copyright to defeat its own ends, stultifying creative practices in a thicket of privately owned algorithmic productions.”[16]

 

This world she describes is not hard to imagine.  A world where AI-generators – at the direction of humans (at least for the time being) – generate anything and everything.  Thus, to continuously expand intellectual property rights merely to encompass the latest technological innovation may bloat the system to the point where it begins to encroach on the public domain – the arena that benefits us all.  To summarize:

 

“Without a principled commitment to the public purposes of copyright, economic interests will surely set our course towards increased commoditization and private control…we may then miss out on the true promise of AI to contribute to a vibrant public domain, instead hindering or harming the very creative processes and cultural exchange that copyright law is meant to foster.”[17]

 

 

 

[1] Andres Guadamuz, Artificial Intelligence and Copyright (October 2017), online: WIPO Magazine Artificial intelligence and copyright (wipo.int)

[2] Théberge v Galerie d’Art du Petit Champlain inc., 2002 SCC 34 at para 30.

[3] CCH Canadian Ltd. v Law Society of Upper Canada, 2004 SCC 13 at para 16.

[4] Geophysical Service Incorporated v Encana Corporation, 2016 ABQB 230 [Geophysical] at para 88.

[5] Thaler v Perlmutter, Case 1:22-cv-01564-BAH (D.D.C., Aug. 18, 2023).

[6] Jerry Kelly, AI-Generated Content Should Be Copyrightable (March 2023), online: Forbes https://www.forbes.com/sites/forbesagencycouncil/2023/03/15/ai-generated-content-should-be-copyrightable/?sh=4b0e94c24ae8

[7] Geophysical, supra note 4 at para 91.

[8] Ibid at para 89.

[9] Ibid at para 91.

[10] Ibid at para 82.

[11] Charlie Warzel, The Most Important Job Skill of This Century (February 2023), online: The Atlantic https://www.theatlantic.com/technology/archive/2023/02/openai-text-models-google-search-engine-bard-chatbot-chatgpt-prompt-writing/672991/

[12] World Economic Forum, 3 new and emerging jobs you can get hired for this year (March 2023), online: World Economic Forum 3 new and emerging jobs you can get hired for in 2023 | World Economic Forum (weforum.org).  See also Tom Huddleston Jr., This is the No. 1 ‘most important’ AI skill you need to know, says MIT expert (September 2023), online: CNBC: https://www.cnbc.com/2023/09/22/tech-expert-top-ai-skill-to-know-learn-the-basics-in-two-hours.html

[13] – ChatGPT3.5, “is prompting AI a skill?” (December 15, 2023), online: OpenAI https://chat.openai.com/

[14] Society of Composers, Authors and Music Publishers of Canada v Entertainment Software Association, 2022 SCC 30 at para 6.

[15] Carys Craig, “The AI-Copyright Challenge: Tech-Neutrality, Authorship, and the Public Interest” (2022), All
Papers. 360.

[16] Ibid at page 22.

[17] Ibid at page 3.