On any given weekend in Hollywood, an assistant is staring down a stack of scripts they’ll never finish.
Script coverage is the first step in a film’s development. A reader summarizes a screenplay, evaluates its strengths and weaknesses, and passes that report up the chain. On one hand, it’s an on–the–fly way to learn structure and develop the elusive instinct for what makes a story worth telling. It is also cheap but time–consuming labor, the kind that quietly sustains this behemoth of an industry.
With thousands of scripts circulating and only a fraction getting read, script coverage is fertile ground for the nascent artificial intelligence revolution.
Kartik Hosanagar, a Wharton professor and founder of the AI script coverage platform ScriptSense, built his company with the goal of streamlining the script coverage process. “The vast majority of screenplays that are sent to agents and producers pretty much go unread,” Hosanagar says. “Even for the things that get read, there’s so much guesswork involved.”
His tool aims to process scripts at scale, summarizing, categorizing, and helping executives decide what to prioritize. After all, coverage takes time, and time costs money. AI is exceptionally good at distillation, able to generate loglines, produce clean synopses, and cross–reference a script against hundreds of others in seconds. For an overworked reader assigned four scripts and two books in a weekend, that kind of speed means survival in an industry as fast–paced and cutthroat as Hollywood.
But efficiency is only part of the story. AI can be remarkably consistent as it summarizes and compares scripts at scale. It can quickly tell a reader what a script is about. It can identify patterns no human reader could reasonably track. What it cannot convincingly do, however, is care.
Hosanagar was explicit about this limitation when designing ScriptSense. The platform would generate anything from summaries to casting suggestions, but would leave blank the most subjective question in any coverage report: What did the script make you feel? It would not recommend whether a script should proceed or pass. That judgement, he insists, must remain human.
“There’s a lot of nuance to reading scripts,” he says. “It’s very easy to prompt an AI system and say, ‘Tell me what it makes people feel.’ It’ll make something up, but it doesn’t mean it’s real.”
Not all AI script coverage programs take such care in preserving the human role, though. Competing AI companies like Greenlight Coverage offer verdicts for those who don’t have the time to do any reading themselves. The gap between information and interpretation is where much of the anxiety around AI in creative industries lives.
Nick Harty (W, E ’27), who’s worked coverage internships before and is interested in pursuing a career in film, sees the stakes in creative terms. “If the best scripts were always getting through, we’d have the worst slate of movies you could possibly imagine,” Harty says. The films that endure are often the “odd balls”—the ones that defy structure, resist easy categorization, or read unpredictably on the page. They don’t have to be the most polished, but the most singular.
AI, by contrast, tends toward the average. It predicts what works based on what has worked. “It just wants to please you,” Harty says. “It’s really not interesting.”
AI’s tendency to identify good scripts through familiar plot lines leads to a greater risk of homogenization within the industry. For writers, the possibility that AI will filter out strange, risky, and unclassifiable scripts feels personal. Aspiring screenwriter Henry Franklin (C ’27) puts it simply: “I would much rather have a human read my script.” To him, the idea that a screenplay might be reduced to a piece of data for a machine to process “doesn’t produce the warmest sensation.”
Despite these anxieties, the logic of the economy continues to dominate. Development is the cheapest phase of filmmaking. But it’s also the easiest to cut. Hosanagar expresses concern that Hollywood studios and production companies will use AI as a tool to replace interns and freelance readers. “With a tool like AI, the worst–case scenario is a lot of these companies get lazy,” Hosanagar says. “I think that is bound to happen.”
Concerns around AI extend beyond efficacy and into industry access. “It’s a genuine concern that the entry point for a lot of people into this industry is impacted,” Hosanagar says. If AI reduces the need for coverage readers, there will be fewer and fewer of the entry–level roles that many young people rely on to break into film.
For decades, coverage has been more than a task—it’s been a rite of passage for aspiring filmmakers. Kaia Chambers (W ’26), who has done coverage for four years across classes, production companies, and talent agencies, describes it as “a training ground.” It’s where young people learn how to read critically, articulate their thoughts, and navigate an industry built as much on relationships as on scripts.
At the same time, access to that pathway has never been equitable. “The ugly truth of it is, even if we all use AI for coverage, studios are never going to stop hiring talented young people from top schools,” Franklin says. “The people who go to Penn will be fine. … Those entry jobs will exist for them.” For those trying to build a career with less connections, though, the outlook is less certain. “The people who get by with their living doing coverage hired by a studio, those are the people who are in trouble,” Franklin says.
As AI begins to absorb the busy work—with programs like ScriptSense already being used at major agencies like WME—the role of the human reader is beginning to shift rather than disappear. In the best–case scenario, something more meaningful remains. Hosanagar hopes that interns can spend less time recounting plots and more time identifying talent, contextualizing projects, or advocating for scripts they believe in. Chambers also notes the shift. Though plot summaries were handled by internal AI systems at her last internship, human readers were still expected to provide the commentary. “I think AI just regurgitates what it thinks it knows. That’s not enough,” Chambers says. “What has impressed [my bosses] the most is if I’m able to have an opinion.”
There’s still a risk embedded in that transition. Coverage, for all its tedium, has always been a volume game. Reading hundreds of scripts builds intuition. Hosanagar says that executives in the industry are worried that relying on AI will erode that foundation, and young professionals “will lose the ability to read a screenplay and say why [it] is interesting and unique.”
Franklin remarks that there are historical parallels to AI’s rise. The transition from physical film editing to digital software was once seen as a disruption. It was faster, more flexible, and ultimately inevitable. But, as Harty notes, the editors who thrived were those who were trained on the old ways and understood the underlying craft. Technology changed the tool, but not the outcome.
The rise of computer–generated imagery also transformed filmmaking, enabling new kinds of spectacle while also inviting overuse and audience fatigue. Chambers says that at a certain point, “it’s not alluring and cool anymore.” Audiences and filmmakers will adapt to the new realities of filmmaking as they always do, she says, as “culture shifts with what’s counterculture.”
AI may follow the same trajectory. As a tool, it can reduce friction, lower costs, and expand access. Hosanagar points to the possibility of a more decentralized industry, where creators can bypass traditional studios and use AI to develop, produce, and distribute content independently. At the same time, studios facing financial pressures may lean into cost cutting, prioritizing speed over nuance. “All of it’s about money,” Chambers says. If reducing a film’s budget is better for the bottom line than gambling on potential box office gains, companies will make those cuts. But audiences, she argues, can tell the difference. “People are realizing who’s using [AI] as a shortcut versus who’s using it in an innovative way.”
That distinction may define the future of AI in the film industry. For students trying to break in, the question of AI looms over everything. Harty says that AI fluency can be a signaling tool—executives may not fully understand the technology, but they are eager to invest in it. “If you can show that you’re adept at using AI to parse through more scripts, but your personal taste is what makes you stand out, that’s going to be huge,” Harty says.
The emphasis on taste recurs across every conversation about the future of film. Coverage, in its ideal form, isn’t about the ability to be correct, but the ability to have conviction. “The best scripts that I read, I wouldn’t even do my coverage,” Chambers recounts. “I’d run into the office and … throw the script on [my] boss’ desk.” AI can simulate feedback, guess at analysis, and even mimic the language of enthusiasm, but it can’t replicate real human urgency.
Despite that, it’s already here. Interns are already using it—sometimes against company policy. Studios and talent agencies are experimenting with internal systems. The industry is in transition, and for students planning on entering Hollywood, AI is less of an existential threat and more an opportunity to rethink the system.
Chambers predicts a counterreaction—a resurgence of independent, countercultural filmmaking that rejects the polish of AI in favor of something more tangible and personal. “Every ten years we hear, ‘Cinema’s dying.’ And it doesn’t,” Chambers says. “It has changed and morphed over the years, but I don’t think that means that AI is going to kill it.”
Franklin says he agreed. “[Some say] all content is going to be made by AI, but I don’t think humans can fundamentally accept that,” he says. “I don’t think people watch things so passively. There will always be value put on human creation.”
With AI seeping into every aspect of human life, the future of cinema isn’t about labor or technology, but what audiences choose to value. If AI handles the first pass of coverage, someone still has to decide what comes next.
That decision, for now, remains human.
