Authors, artists, and their employers are furious over generative artificial intelligence (AI) – chatbots that write novels and news copy, image generators that create artwork or music to order in the style of any artist whose work is accessible on the internet. The list of pending lawsuits is long and growing.
Major media outlets including The New York Times and The Chicago Tribune contend that their stories were copied “with impunity,” prominent authors of fiction and nonfiction allege “systematic theft on a mass scale,” and famous artists assert that image generators duplicate their work and threaten to destroy their livelihoods.
The authors and artists object not only to the actual generation of prose, images, or music by AI but also to the use of their work to train the AI in the first place. Reading some of the pleadings, you sense genuine anxiety in addition to grievance – the feeling that the more prolific and successful the plaintiffs become, the faster they are training their replacements.
Their fears are borne out by well-publicized waves of AI-driven layoffs across the entertainment industry and even among the coders themselves. The legal questions, however, are complex and won’t be settled for years. They involve the reach of copyright and the limiting doctrine of “fair use,” as well as the licensing terms that content creators impose on consumers of their work.
Resolving these questions is easier, or at least clearer if we are prepared to attribute agency to a computer and judge its activities as if they were undertaken by humans. At one level, of course, this is ludicrous. Machines don’t think or create like humans – they just do what we tell them to do.
Until very recently, it was easy to see computers as sophisticated tools subservient to human agency, regurgitating pre-loaded content and crunching numbers. Today, we converse with chatbots the way we would with a research or coding assistant, and with image generators the way art directors guide human illustrators and graphic designers.
Much as it discomfits us, generative AI learns and, at some level, “thinks.” Trained on a significant slice of human knowledge, ChatGPT aced the “Turing test” – the famous measure of a machine’s ability to exhibit human-like intelligent behavior – the day it was released.
Since then, chatbots have passed the bar and medical licensing exams, solved long-standing math conundrums, and written more empathetic responses to patient questions than their doctors. They even outperform humans on tests of creativity, and it is precisely to encourage creativity that copyright laws exist.
This is not to argue that we should start according rights to machines – far from it. Rather, it is because humans develop and benefit from generative AI that we must ask if there is any legal basis for treating AI differently under copyright law. Humans read books and newspapers to learn, to become more informed, and to become better writers. No one argues that violates copyright.
They may bring sketchbooks to museums and record their impressions of works they see, improving artistic skills and broadening stylistic repertoires. All agree this is “fair use.”
Why, then, should training a generative AI on publicly accessible content be prohibited? AI systems may not learn or think like biological beings, but learn they do, and, whether or not we choose to call their inferential processes “thinking,” they clearly exhibit intelligent behavior.
Leave questions of ontology and the roots of knowledge to the philosophers (or, if you prefer, to their imitative chatbots). We benefit humankind by making the rudiments of knowledge creation accessible to generative AI. AI helps us perform better. It juices our game, as long as we don’t forget how to think for ourselves.
We expect doctors to keep up with the medical literature and lawyers to read the latest cases, so if we value the assistance AI provides, we should want to see it exposed to the broadest possible swath of human understanding. It is hard to see how this violates anyone’s rights.
The copyright plaintiffs have another theory: that the trained chatbot is so directly derived from their proprietary content as to constitute an infringing work in itself. But chatbots are based on “large language models,” which organize massive numbers of basic text elements into a complex representation that captures meaning and word relationships.
This allows the chatbot to formulate coherent answers to queries. It seems difficult to argue that such an esoteric representation, based on so many written works, is any more infringing than a human brain exposed to The New York Times.
Whether generative AI, once trained, might produce stories or images that infringe on someone’s copyright is a completely separate question. In its lawsuit, The New York Times cited instances of verbatim copying of its content by ChatGPT. Depending on how much was copied, those specific instances could represent copyright infringement regardless of whether the culprit is man or machine. (OpenAI, the proprietor of ChatGPT, insists such cases are rare and thinks their chatbot may have been tricked into copying.)
Artists have a tougher case to make because style has never been protectable by copyright. Today, anyone is free to hire an artist to create a work in the style of another artist. That may be crass but, as long as no specific work by the other artist is copied, it isn’t legally actionable. Why should we hold image generators to a different standard?
While the artist side of the story pits individuals against corporate behemoths like Microsoft, the chatbot wars are a clash of titans: the plaintiffs are Big Media and wealthy authors. They’ve sought to bolster their copyright case with terms of service that prohibit scraping content for use in AI training. Unfortunately for the plaintiffs, you can’t enlarge copyright using contractual restrictions.
If the use of your content is either noninfringing or falls within “fair use,” which places activities such as research and teaching beyond the reach of copyright, those restrictions are unenforceable.
Like all technology revolutions, the advent of generative AI will produce winners and losers, but potentially on a vaster scale – and with a greater share of losers – than any previous advance due to the sheer number of jobs it affects. No career is truly safe because the imperative to save money is universal.
Of course, it’s easy to say that technology has been coming for jobs since the Industrial Revolution and the net effect has always been more jobs, not fewer – until it’s your job. But the galloping popularity of generative AI attests to the undeniable and widespread benefits it delivers. Fettering it with legal hopples that benefit one set of big players over another will not, in the long run, and probably much sooner than that, reduce the inevitable dislocation.