I experimented with Dramatron from DeepMind.

  • Like
    • Hyerarchical approach of starting from a logline, to outline, to writing individual scenes
    • Can be run from a Notebook (once you have Vertex AI running - the whole process made me appreciate the focused approach OpenAI has for ChatGPT)
  • Potential improvements
    • Do not rely on summaries of previous scenes - While summarization is a strong suit of LLMs, I believe that summarization is not a core part of the writing process. Ultimately, a screenplay is about characters pursuing their goals in the immediate scene. They aren’t summarizing their life story in their head; they are acting and reacting. As a writer, we may use summarization to pick the scenes to include, but that is more equivalent to picking from hundreds if not thousands of possible scenes. Feeding a scene summary to an LLM is asking it to pick the next probabilistically, it is taking one of the most powerful tools of a screenwriter and reducing it to the style found in a children’s book. If an LLM is going to write a great screenplay, it is going to have to do more than summarize and continue (which probably points to issues with an LLM outlining a screenplay as well)
    • Do not rely on a theme - Similar to relying on summaries but I can lean on David Mamet in his Master Class on dramatic writing holds up a script, shakes it, and says no theme falls out of it. The only thing in a screenplay is the characters taking action in pursuit of their goals. I’m not saying that there isn’t a theme, but we need to be careful not to conflate the appreciation and study of art for how art is created.

Generative AI Industry

  • Attened The AI Conference which was awesome however many of us couldn’t escape the feeling of “Selling pickaxes during a gold rush
  • Also got to Google Cloud - Generative AI Live + Labs Sunnyvale
    • “LLMs are like fancy autocomplete”
    • “Hottest programming language is English”
    • Google sticking it’s AI efforts into the convoluted soup that is Google Cloud is frustrating and in poor contrast with the focus and simplicity of OpenAI. Is it me or has everyone working in the Cloud just come to expect clutter and lack of clear vision? Or am I off base for not being their target audience.

Use Cases that just might be viable

  • None Player Characters (NPCs) - The value add here is pretty clear and a few startups have picked up on this opportunity. The challenge, however, lies in transcending the status of a mere LLM wrapper. What unique value can these startups offer that one can’t achieve by simply crafting a well-thought-out prompt for an LLM?
    • The value might also prove to be more a novelty. This brings to mind the perennial debate in screenwriting: what’s more critical, character or plot? I approach this conundrum through the lens of risk, especially considering the cardinal rule of the craft: “Never be boring.” Consider Quentin Tarantino’s style, where minutes of screen time are devoted to punchy, character-driven dialogue. While this can be engrossing, it’s also a gamble. If the subject matter doesn’t resonate, the pacing stumbles, veering into tedium. In a series vs a movie, characters are given ample room to “breathe,” but this extended focus can border on the monotonous. Returning to the example of “The Godfather,” the film manages to engage nearly everyone, despite its focus on otherwise polarizing Mob characters. When you hinge your narrative solely on characters, you skate on thin ice, jeopardizing your audience’s interest. I believe the same will prove true with LLM powered NPCs over time.