Notes

← Back to home

A collection of fragments of understanding in the pursuit of deeper questions.

AI & Creativity

A more honest title... Copyright Authorship and Machine Learning Behind the AI veil - What is Artificial Intelligence? Turing test - evaluate a machine's ability to appear human (Alan Turing, 1950). "Machines that are capable of performing tasks that, if performed by a human, would be said to require intelligence" (Scherer, 2016). "Systems that display intelligent behaviour by analysing their environment and taking actions -- with some degree of autonomy -- to achieve specific goals." (EU Commission, 2018). (Preferred definition).

Artificial Intelligence - Machine Learning Autonomous machines? Machine learning - not new, yet...

  • Growth in computing power
  • Data availability
  • Progress in algorithms

Supervised vs Unsupervised learning Neural networking - learn from example

Machine Learning

  • Large Datasets, accuracy
  • Large computational power, speed
  • Advanced Algorithms, precision

It is the ability to extremely fast make sense of input data with high accuracy without being explicitly programmed to do so.

What is Artificial Intelligence? From an IP perspective it is useful to distinguish:

  • Assisted
  • Augmented
  • Autonomous (= "Computational")

Recent literature on artificial intelligence distinguishes three different forms, namely assisted, augmented and autonomous artificial intelligence. The difference between these forms comes from the allocation of decision rights between man and machine. In the assisted artificial intelligence stage, machines execute certain specific tasks, but decision rights remain solely with human beings; in the second stage, augmented artificial intelligence, humans and machines share decision rights and learn from each other; and in the third and final stage, autonomous artificial intelligence, machines ultimately take over all decision rights, either because humans increasingly trust the machines' abilities to decide, or because decisions have to be taken so quickly or require so many data that humans are simply unable to decide.

EU Copyright Law Implications Re-cap - Basics of Copyright Law

  • A bundle of rights granted to the author of an original work for a limited period of time starting from the expression of the work.
  • Legislation framework - very little harmonization in Europe - 11 Directives and 2 Regulations.
  • Duration - 70 years PMA.
  • Subsistence - Originality
  • Rights - Economic and Moral
  • Remedies - Injunction, damages, account of profits, etc.

Authorship - a two-fold analysis

  • Designation issue - who is the author and, in the context of machine learning, can authorship of copyright be ascribed to a non-living person?
  • Originality issue - can an AI-generated work be protected as original under EU law?

Designation - Can authorship of copyright be ascribed to a non-living person? No definition of an "author" in the law, however Human authorship implied - anthropocentric view

  • Within copyright law framework
  • Traditional justificatory theories:
    • Utilitarian theory
    • Locke Labor theory
    • Personality/Natural rights theory
  • According both to Copyright Law Framework and Traditional Justificatory theories, the human is at the center, so Human Authorship.

Authorship in the Law

  • Art. 6bis Berne Convention

    "Independently of the author's economic rights [...] the author shall have the right to claim authorship of the work and to object to any distortion, mutilation or other modification of, or other derogatory action in relation to, the said work, which would be prejudicial to his honor or reputation. The [moral] rights [...] shall, after his death, be maintained, at least until the expiry of the economic rights, and shall be exercisable by the persons or institutions authorized by the legislation of the country where protection is claimed.

  • Art. 1 Term of Protection Directive

    The rights of an author of a literary or artistic work within the meaning of Article 2 of the Berne Convention shall run for the life of the author and for 70 years after his death, irrespective of the date when the work is lawfully made available to the public.

Work protected by Copyright Originality - Can an AI-generated work be protected as original under EU law? For long no uniform understanding of originality - classical common law vs civil law opposition. Infopaq[2009], "Author's own intellectual creation". Yet, no definition of "creation". Circular definition.

Originality - In the Law

  • Art. 1(3) Software Directive, Art. 3(1) Database Directive, Art. 6 Term of Protection Directive

    Computer programs, databases and photographs shall be protected if they are original in the sense that they are the author's own intellectual creation. No other criteria shall be applied to determine their eligibility for protection.

  • Recital 16 of Term of Protection Directive

    A photograph work [...] is to be considered original if it I the author's own intellectual creation reflecting his personality.

Originality - In the Case Law

  • CJEU -- Infopaq decision

    Extension the notion of originality as author's own intellectual creation to subject-matter falling within InfoSoc Directive.

  • CJEU - Football Dataco decision

    *[T]he notion of the author's own intellectual creation refers to the criterion of originality. [T]hat criterion of originality is satisfied when [...] its author expresses his creative ability in an original manner by making free and creative choices [...] and thus stamps his "personal touch". By contrast, that criterion is not satisfied when the setting up of the database is dictated by technical considerations, rules or constraints which leave no room for creative freedom.*

US approach to Authorship A discusses in Section 306, the Copyright Act protects "original works of authorship". To qualify as a work of authorship" a work must be created by a human being... The Office will not register works produced by nature, animals, or plants. Likewise, the Office cannot register a work purportedly created by divine or supernatural beings, although the Office may register a work where the application or the deposit copy(ies) state that the work was inspired by a divine spirit. Examples:

  • A photograph taken by a monkey.
  • A mural painted by an elephant.
  • A claim based on driftwood that has been shaped and smoothed by the ocean.
  • A claim based on cut marks, defects, and other qualities found in natural stone.

Similarly, the Office will not register works produced by a machine or mere mechanical process that operates randomly or automatically without any creative input or intervention from a human author.

US approach is less anthropocentric than EU.

image64

Practical Problems

  • Certain interests to be incentivized
  • Producing AI-generated works benefits society generally.
  • Undermining human authorship - impact negatively the market for low creativity works.

There is the necessity to adopt a practical solution that "makes sense in terms of doctrine, but also in terms of realities".

Some of the potential authorship claims

  • The user of the device, can someone who clicks a single button without determining any of the creative output and as such without employing intellectual creation be considered an author (patterns established by ML algorithm determine the creative output)?
  • The programmer
    • It's for the programmer's creativity in the first place that the final output comes into existence.
    • However, no longer coded rules to follow, but examples to learn from...
    • The programmer creates the potentiality for the creation of the output, but not its actuality".
  • The device (AI)
    • Human Authorship!
    • No need to be incentivized.
    • Absence of personality.
    • Lacks ability to exercise intellectual labour.
  • Joint authorship
    • Absence of intention to be joint authors.
    • Fractioning the authorship claim.
    • Little harmony of interests between programmer and user.

In search for a pragmatic solution - Common Law Concepts - legal fictions

  • UK - Copyright Designs and Patents Act (CDPA) 1988.

    Section 178: "computer-generated work means that a work is generated by computer in circumstances such that there is no human author". Section 9(3): "In the case of a literary, dramatic, musical or artistic work which is computer-generated, the author shall be taken to be the person by whom the arrangements necessary for the creation of the work are undertaken".

  • A Straightforward Case

    In a British case concerning the frame images generated when the user plays the game, the person by whom arrangements were undertaken was considered to be the person who programmed and designed the game. The Court expressly refused to confer that status to the user, as the latter's input was not artistic in nature, nor had he undertaken the arrangements necessary to create the frame images.

  • Less Straightforward Cases and the Issue of Legal Uncertainty

    In short, the uncertainty regarding the person by whom arrangements are undertaken, who will have to be identified on a case-by-case basis, does not favor legal certainty and constitutes a reason to not extend the applicability of this legal fiction.

  • US - Works Made For Hire Doctrine.
    • Section 201 US Copyright Act 1976

      "In the case of a work made for hire, the employer or other person for whom the work was prepared is considered the author or purposes of this title, and, unless the parties have expressly agreed otherwise in a written instrument signed by them, owns all of the rights comprised in the copyright."

  • A mismatch between the author-in-fact (employee or contractor) and the author-in-law (the employer or another person who has paid for the work).
  • "Deemed" authorship through a contract/payment.
  • And Ai generated works? The relationship between the creator of the AIs and the Ais does not fit the characterization as commissioner-creator or employer-employee (not least because the employee or agent has legal rights and duties by agreement with the employer/commissioner).

Some conclusions

  • Many creative works as a result of machine learning.
  • EU Copyright System (as well as US) - the human author is central.
  • Practical problems if public domain solution adopted - low creative works.
  • Potential authorship claims not working.

Other solutions...

  • Sui generis regime?
    • Drawing parallels with the database producer's right, the most suitable rationale for protection of AI generated works would be investment protection.
  • Public domain solution with publication/dissemination rights for a limited amount of time.
  • Short term copyright protection with no moral rights.
  • The right solution to be adopted needs economic justification -need to understand whether market failure exists: this should be assessed together with economists.
  • Solution will have to be coordinated with liability for AI actions.