Beyond Modicum Of Creativity: Human Agency And Future Of Copyright Authorship

Nonetheless, copyright law provides the requirement of minimum creativity also known as 'modicum of creativity'. The courts have set this creativity requirement at a low pedestal that means even slightest creative work would qualify for protection. But scholars have repeatedly described this minimal requirement of creativity as 'substantively impotent', and consider it as having little to no use in copyright analysis. It has been emphasized that the requirement of creativity has to be found in the decisions and choices made by the author, manifested in the created work rather than the creative process. Now, supposedly generative AI produces some content after processing on trained data, it might qualify the requirement of originality if there is minimal creativity. So, theoretically generative AI might qualify for authorship (considering other aspects as well). But scholars have emphasized that these generative AI models are purposefully trained to learn and reproduce expressions like tone, style, and structure, and this is not an 'incidental' processing of data but a designed tool to generate similar content. It was also argued that generative AI models are not some magical portal which form fresh information from some different universe but it is a data structure that consists of information from its training data.

As the models of generative AI trained on large-scale databases, which often includes millions of copyrighted works, these models function by anatomizing the patterns and predicting the most probable words or pixel according to their training data. Thus, the output is a probabilistic rearrangement of the existing works rather than any original creation. If a requirement of originality is lowered to include algorithmic or automated outputs, the 'public domain' would be inundated with AI-generated content, and this would lead to monopolization of patterns and styles which were originally contributed by humans.

Now, even if we do not take into account some of the abovementioned six resources and assume generative AI as an author (on the basis of modicum of creativity), then does its content fulfill the criteria of having an intended audience for it to have certain social value? Because the requirement of appropriateness is also something which could not be ignored. It tracks whether a work has some social value. As René van Woudenberg et al., gave emphasis to the point that we as humans always have a certain audience to target for writings, and it requires a 'consciousness'. On the other hand, AI systems like ChatGPT have no targets, it produces texts according to certain prompts [it is more reactive in nature (prompt-and-generate)]. The CEO of OpenAI has a broad aim for it to have users but ChatGPT itself does not. It might display intentionality but they are not the result of intentional writing. Woundenberg has considered this intentional writing as normal writing in which (i) author has an intention to write, (ii) the author knows what they are writing, (iii) author targets a readership, (iv) author has intentionality in writing; (v) author has intention to produce some effect on the readers, and lastly (iv) author writes 'his own words'. It was emphasized that the alleged requirement of intentionality in the case of generative AI could be understood by the theory of connectionism, which explains processing of information as interactions between neurons within neural networks. Such systems rely on external inputs as it lacks sensory abilities and it does not see, smell, touch, hear, or feel. It is needed to understand, AI models could not comprehend the context like humans do because of lack of deep semantic understanding. That means, there is a significant gap between creation of new ideas and generation of new text.

Unlike originality and creativity, the 'targeted audience' or 'consciousness' is not the absolute requirement for a protection but it represents the work having human agency which brings truthfulness (which is not a design principle of generative AI), and capability of taking responsibility for any wrongdoing. Therefore, the author is of the opinion that the discussion over the years on whether generative AI could be considered as an author under copyright law is not a viable question to ask at this juncture. Reason being, its premise is coming from the assumption that a thing which generates any work through its own way could be considered as 'original' if it qualifies the minimum creativity requirement. It is important to remember that human-centric authorship is deep rooted in the natural right of a person to the expression of their brain, and by maintaining a human requirement for originality, copyright law incentivizes human ingenuity rather than a mechanical algorithmic data-mining. By looking into the working of generative AI, it 'reproduces' the work of already existing works and that work comes into picture only when trained using millions of data. While arguments on minimal requirements of creativity would somehow fulfill the criteria of protection and label generative AI as an author, yet the significant gap remains, and the answers to various questions of AI authorship which could only be addressed by defying the rights of human beings or considering AI as a human.