From ChatGPT experiments to WGA contract protections, how AI is and is not changing screenwriting

The Reality Behind the Headlines

LOS ANGELES — The 2023 WGA strike produced contract language on AI use in film and television that the Guild described as “unprecedented protections” and that studios, notably, agreed to without the resistance some expected. The contract provisions prohibit studios from using AI-generated material as a writing sample or source material in ways that would undercut WGA minimums, and require disclosure when AI-generated material is provided to a writer, while preserving studios’ right to ask writers to use AI tools and writers’ right to refuse.

What has happened in practice is more nuanced than either the apocalyptic or the dismissive view of AI’s screenwriting impact would suggest. Writers report that current AI tools are useful for certain specific tasks — generating alternative lines of dialogue to consider, producing scene outlines from a writer’s detailed notes, identifying structural patterns in competing scripts — and largely inadequate for the core tasks of screenwriting: generating emotionally true character behaviour, constructing dramatically effective scene dynamics, and producing the unexpected specific detail that distinguishes good writing from competent writing.

The Tasks AI Does and Does Not Do

AI language models trained on large corpora of screenplays can produce output that follows screenplay format, includes the conventional elements of a scene heading and action block, and contains dialogue that sounds plausible at a sentence level. They cannot, currently, produce output that experienced writers describe as good — writing that surprises, that captures specific human truth, that demonstrates the authorial intelligence that makes the difference between a produced script and a passed one.

This is both a current limitation and, writers argue, a likely permanent one: the specific and surprising detail that distinguishes excellent writing requires a relationship to experience, to subjectivity, and to the emotional intelligence that human writers bring and that current AI systems demonstrate only in its absence — in the generic, the averaged, the statistically probable phrase where the writer’s instinct would choose the precise and unexpected one.

The Business Pressure

The business pressure to reduce writing costs through AI is real and will not disappear because writers find the tools inadequate for their best work. Studios and streaming platforms will experiment with AI in development, in coverage, and in lower-tier content production. The Deadline AI coverage documents specific announcements as they occur. The WGA contract provides a framework. The enforcement of that framework, and the creative quality of content produced under various AI-assisted conditions, will determine whether the protections hold under the commercial pressure they were designed to resist.

SOURCE: https://bohiney.com