E30: The AI Revolution in Hollywood? With WGA writers Trey Kollmer, Sophia Lear, and Garrett Schabb

Episode Summary

Executive Summary: This episode examines the WGA strike through the lens of AI’s impact on Hollywood writing. The guests—three working writers—generally welcome AI as a tool but fear studios using it to shrink rooms, devalue writers, and automate cultural production. They debate creative quality, enforcement, copyright, residuals, training data, and whether AI will preserve or erode shared human storytelling.

Main Topics: WGA strike demands and AI restrictions (Priority: 5/5): The writers explain the guild’s core AI asks: no AI-written scripts, no AI rewrites, no AI-generated source material, and no use of guild-covered material to train models. They frame these demands as protections against displacement and ownership theft. How AI currently fits into writers’ rooms (Priority: 5/5): The guests describe limited but growing experimentation with ChatGPT/GPT-4 for outlining, brainstorming, structuring scenes, and organizing ideas. They generally find it useful for medium-level scaffolding but weak at specificity, originality, and jokes. Job design, staffing, and gig-ification fears (Priority: 5/5): A major concern is that AI could let a few showrunners and one or two writers generate drafts while the rest of the room becomes cheap polishers. The strike also seeks minimum staffing, especially as streaming rooms have already shrunk. Creative quality, originality, and cultural value (Priority: 4/5): The writers argue that the best TV and film are often polarizing, human, and shaped by room alchemy. They fear models will default to clichés, encourage bland consensus, and produce endless regurgitation rather than new voices. Curation, personalization, and future entertainment formats (Priority: 4/5): The conversation explores whether AI may shift value toward curators and tastemakers, and whether future media will be personalized, interactive, or generated on demand. The writers worry this could weaken shared culture even if it increases choice. Labor, social contract, and broader societal reform (Priority: 4/5): The host and guests discuss whether AI requires new protections such as UBI or structural reforms. Writers see their strike as part of a larger labor response to automation and note that less-protected workers will have fewer defenses. Training data, copyright, and creator rights (Priority: 4/5): The group raises unresolved questions about using famous writers’ styles, studio-owned scripts as training data, and whether creators should share in upside if their work helps models. They emphasize the legal and ethical gray areas remain unsettled.

Key Arguments: Writers are not Luddites; they are curious about AI and often use it as a tool, but they want to preserve human authorship and fair pay. AI is already helpful for outlines, scene structure, brainstorming, and prompt-based note-taking, especially for newer writers or narrow tasks. The biggest labor risk is not full replacement tomorrow, but gradual deskilling and a gig-economy model where AI does first drafts and humans only punch up. The WGA’s minimum staffing and AI restrictions are linked: keeping more writers in the room helps prevent AI-driven hollowing-out of the middle of the profession. Current model behavior tends to be vague, cliché, and over-sanitized, which may reduce the edge and polarization that make great comedy and drama work. There is no simple policy answer because AI’s effects differ by domain: medicine and law may benefit from clear standards of care, while culture raises harder questions about authenticity and social meaning. Enforcement is difficult because AI outputs are hard to detect reliably, and superficial edits can bypass detectors, making contractual rules necessary but imperfect. If AI gets much cheaper and more capable, entertainment could become personalized and abundant, but shared cultural experiences may diminish. A positive utopian future would separate basic survival from work, allowing people to create for meaning rather than wages, but speakers doubt society will reach that point smoothly. Studios may prioritize cost savings over quality, especially if AI lets them reduce staffing or ownership obligations; writers see this as a threat to both livelihoods and storytelling standards.

Data Points: Writers Guild strike-related interviews: 3 - The episode centers on three WGA members: Trey Colmer, Sophia Lear, and Garrett Schaub. Years in the WGA for Garrett Schaub: since 2012 - Garrett describes his tenure in the Writers Guild. Years writing for Trey Colmer: since about 2010 - Trey says he has been working as a TV writer for roughly this long. Writer’s room size on Trey’s current show: 14 writers - Trey notes the staff size on Ghosts. Typical network TV season length: 22-episode - Sophia describes traditional network shows as 22-episode productions. Writer’s room duration on network shows: 9 months - Sophia says network shows involve about nine months of room work per season. ChatGPT availability before strike: 3 weeks - Trey notes they were only back for three weeks before the strike hit. Pilot scripts once bought by networks in the old system: about 70 - Trey describes the historical network development pipeline. Pilot productions typically chosen: 10 to 12 - Trey explains how many bought pilots were actually filmed. Greenlit shows in the old system: 4 or 5 - Trey estimates how many pilot contenders became real series. Model preference gap: 70 to 30 - The host cites a comparison where GPT-4 is preferred over GPT-3.5 by roughly a 70/30 ratio. Performance jump referenced: bottom 10% to top 10% - The host compares GPT-4’s improvement to a dramatic leap on the bar exam. Training cost decline: 97% price reduction - The host says equivalent model pricing dropped dramatically over about eight months.

Pivotal Quotes: "“The big fear is that if these models are kind of a force multiplier... most of the writers become these kind of gig economy cheap writers who come in and polish it and punch it up.”" — Trey Colmer: Trey explains the central labor concern: AI could hollow out the profession by reducing writers to low-paid finishers. "“If AI can really do a writer’s job as well or better than humans, I don’t think that there’s anything that we can sort of put into contractual language or kind of negotiate to change.”" — Sophia Lear: Sophia reflects the view that the long-term issue is capability itself, not just contract language. "“We’re trying to defend the sanctity of really good storytelling.”" — Garrett Schaub: Garrett frames the strike as both economic defense and cultural stewardship.

Implications: The episode suggests AI may quickly become a writing assistant, but the real fight is over who controls culture, credit, and compensation. For Hollywood, the next few years may determine whether AI augments human creativity or normalizes cheaper, smaller, more automated production.