‘Almost unbelievably good’? The measure of (artificial) intelligence in a changing world for screenwriting

Abstract

2023 was the year AI took centre stage. Beyond the wider public debate, the EU AI Act will now enforce transparency requirements (i.e. disclosure of AI-generated content and of copyrighted data used to train generative AI) and that AI systems be overseen by humans. While clearly informed by a philosophy that wants AI to be useful to people rather than replace them, the act also identifies ‘high risk’ areas for employment, thus echoing the findings of a recent OpenAI working paper assessing the impact of large language models (such as ChatGPT-4) on labour and projecting ‘high’ or ‘full exposure’ for professions such as writers, creative writers, authors, journalists, interpreters, translators, editors and copywriters. In the United States, the long Writers Guild of America’s (WGA) strike culminated in a new Minimum Basic Agreement that represents a landmark precedent aiming to affect labour relations in the writers’ favour. It also introduces transparency requirements on the use of AI by companies and, crucially, it establishes that AI-generated material cannot be considered literary, source or assigned material. Furthermore, the guild ‘reserves the right to assert that exploitation of writers’ material to train AI is prohibited’ by law. Meanwhile, back in Europe, the Writers Guild of Great Britain (WGGB) has followed up with much milder ‘recommendations’ and 2024 ended with Content London launching its first AI Festival Screenings, showcasing twelve AI-generated films in the programmatic hope that ‘the business gets to grips with generative AI’. This fast-evolving scenario begs the hard question: can/will AI replace screenwriters and, for that matter, other professions working with scripts (e.g. readers, editors, etc.)? This article proposes a preliminary assessment of the use of AI in screenwriting. From multiple pulpits we are assured that tools such as ChatGPT can generate ideas, outlines and other development documents and plotlines, enhance character development and improve dialogue, among many other things. Can they really? By combining a genetic approach with close analysis of the original development documents of The Last Screenwriter (2024) – promoted as the first-ever feature film ‘written entirely by ChatGPT-4’ – the article highlights shortcomings as well as raises questions about the potential shifting of screenwriting skills towards a form of prompt engineering.

Ähnliche Arbeiten