Govind Kumar Chaturvedi is an IPilogue Writer and an LLM graduate from Osgoode Hall Law School.
We sat down to chat about how he registered Suryast in Canada. Mr. Sahni told me that he had been inspired by Ryan Abbott’s DABUS, to take on this intellectual property legal experiment. I wanted to learn more about his A.I. and his legal reasoning.
RAGHAV: The A.I.
Ankit shared that his A.I. tool was named “Raghav’. A team of software developers and had gotten the A.I. assigned to him. Raghav’s unique way of working was based on a technique called Neural artistic style transfer, which is inspired by the biological neurons of the nervous system. Just like in the nervous system, the neuron takes in several incoming signals and creates a resulting signal from the inputs. Similarly, an artificial neuron takes input and many artificial neurons form a layer called the neural network. The input can be text, descriptive values, etc. and the output layer can be a label predicting a category like a ‘dog’ or ‘house.’ The user then sees two columns, allowing users to input the image’s style and content. In this case, Sahni chose the Starry Night of Van Gogh for Suryast. The A.I. was already trained on different painters’ data sets. This data set was used that to make the new image and the A.I. was advanced enough to know where to place colours and structures in the painting to mimic Van Gogh’s original work.
Legal Reasoning for Co-Authorship
According to Sahni, Raghav chooses and creates the brush strokes and colour palette, blurring the lines separating his own contributions. Sahni contributed the style and inputs, so the final product is a mixture of both his and Raghav’s work.
I was intrigued about whether A.I. could be considered an author according to the laws of Canada. Currently, the Copyright Act is silent on the issue. Jurisprudence in cases like Setana Sport Limited v 2049630 Ontario Inchas stated that non-juristic persons cannot be authors as the authors have lifetime and must be human. However, by co-authoring Suryast with the AI, Sahni met the legal recommendations for authorship, as it was an AI-assisted work. His creativity and skill were also present in the final work of Suryast and like he said no line could be drawn between his contribution and that of the AI, so the same qualified for copyright protection. I recalled the Copyright Act recognises joint ownership of work under section 2 as work of joint authorship, defined as a work produced by the collaboration of two or more authors in which the contribution of one author is not distinct from the contribution of the other author or authors. As Raghav contributed its own creativity, it fulfilled the definition of joint authorship under section 2.
A.I. is More Than Just a Tool
When asked if AI is just a tool, Sahni re-affirmed that the AI chose how to apply the data set fed to it, suggesting that it was more than a tool. Sahni believed that this contribution met the threshold of minimum amount of creativity required and cited the American case Home Legend LLC v. Mannignton Mills Inc to support this point. In that case, the defendant’s selection and creative co-ordination of images was found to meet the threshold of minimal creativity as the artistic judgment was exercised. Further, in Feist Publications, Inc., v. Rural Telephone Service Co., para 44 states that “As discussed earlier, however, the originality requirement is not particularly stringent. A compiler may settle upon a selection or arrangement that others have used; novelty is not required”. The judge continues at para53 “It is equally true, however, that the selection and arrangement of facts cannot be so mechanical or routine as to require no creativity whatsoever. The standard of originality is low, but it does exist.” Therefore, Sahni believes that human inputs exceed the minimum recognized originality prescribed by law by the Supreme Court of the United States of America. However, while Sahni was able to register Raghav as author, his ownership of Raghav is also an important factor, and authors who do not own their AI co-author may not be as successful.