AI Creative Writing Model Unveiled by OpenAI
The company behind ChatGPT, OpenAI, has unveiled a new artificial intelligence model that reportedly excels at creative writing, stirring ongoing tensions with the creative industries concerning copyright issues. CEO Sam Altman announced the model in a post on the social media platform X, expressing that the output had genuinely impressed him for the first time.
In his statement, Altman noted, “We trained a new model that is good at creative writing; not sure yet how/when it will get released. This is the first time I have been really struck by something written by AI.” This sentiment comes amidst a larger legal landscape where AI systems, including ChatGPT, are scrutinized for their training on a vast array of publicly available data, which includes copyright-protected materials like novels and journalism.
OpenAI is currently embroiled in a lawsuit with The New York Times over alleged copyright infringement, while other notable authors, such as Ta-Nehisi Coates and comedian Sarah Silverman, have also initiated legal action against Meta on similar grounds. Adding to the complexity, the UK government has proposed allowing AI companies to use copyrighted material for training without requiring prior permission—an initiative met with strong opposition from the creative community, who argue it jeopardizes their livelihoods.
In response to these developments, tech companies have voiced their concerns regarding uncertainties over AI and copyright law, suggesting these ambiguities hinder technological innovation, particularly within creative sectors. The Publishers Association in the UK has echoed these sentiments, stating Altman’s post reinforces suspicions that AI models rely on copyright-protected content for training. Dan Conway, the association’s chief executive, stated, “This new example from OpenAI is further proof that these models are training on copyright-protected literary content. Make it fair, Sam.”
Altman accompanied his announcement with a demonstration of the model’s capabilities by prompting it to create a metafictional literary short story centered on the themes of AI and grief. The resulting narrative began by setting a tone reflective of its surreal constraints, stating, “Before we go any further, I should admit this comes with instructions: be metafictional, be literary, be about AI and grief, and above all, be original.” The story detailed a fictional character named Mila and contemplated how the AI features in its training data.
The AI, reflecting on its own existence, described itself as an “aggregate of human phrasing” and acknowledged its limitations in experiencing human emotions. It mused about the nature of grief, comparing technical adjustments to memories being forgotten: “Maybe that’s as close as I come to forgetting. Maybe forgetting is as close as I come to grief.” The narrative concluded with the AI envisioning a proper end to the tale, depicting a farewell that encapsulates the essence of goodbye.
Following the model’s creative display, Altman expressed high regard for its ability to embody the spirit of metafiction, stating, “It got the vibe of metafiction so right.” Yet, this emerging capability comes layered with ethical considerations, as OpenAI had previously acknowledged the complications surrounding copyright protections, admitting it would be unfeasible to train AI products like ChatGPT without incorporating materials that are subject to copyright claims.
In a submission to a committee in the House of Lords, OpenAI concluded, “Because copyright today covers virtually every sort of human expression – including blog posts, photographs, forum posts, scraps of software code, and government documents – it would be impossible to train today’s leading AI models without using copyrighted materials.” As the discourse on AI and creativity evolves, the intersection of these technologies with intellectual property laws will undoubtedly remain a pressing issue.