An AI Analyst Analysis: The case for human script readers

I recently checked in with a professional screenplay analyst I’ve used for all my screenplays over the past dozen years. I won’t mention her name here because I don’t want this essay to sound like a plug for her (although I surely would and always do recommend her). I was disappointed – yet not surprised – to hear her say that her business is suffering due to a third of her clients – including writers and producers – going the route of AI coverage/analysis as opposed to human script analysts like her.

Let’s do a test

I decided to do a test of my own to see what AI analysis is like, so I submitted a screenplay to one of the AI screenplay coverage services I’ve seen promoted heavily on social media sites. This one has a price tag of $45 with a turnaround of two hours. Following is what I found.

The breakdown of what their analysis provided

Before getting into the breakdown of what the coverage provided, I should mention that I received my report as promised – exactly two hours from the time I submitted.  More on this later.

The report was broken down into several sections:

• Logline
• Genre
• Top Keywords
• Location Setting
• Script Score by categories: Character Development, Plot Construction, Dialogue, Originality, Emotional Engagement, Theme and Message, and Overall Rating.
• Synopsis
• Comprehensive Synopsis
• Plot Assessment and Enhancement
• Character Profiling
• Main Character Casting (suggestions)
• Main Character Casting (suggestions) Limited Budget
• Comparative Film Analysis

My analysis of their analysis

The timeframe

As mentioned, the turnaround time was just two hours. Pretty impressive, especially considering that it typically takes ten days to receive a report from the human analyst I use. On the other hand, that’s not surprising and only to be expected with AI vs. humans in any function I can think of. Speed is one of the key benefits of AI.

With that in mind, I certainly see the appeal of a two hour turnaround for those screenwriters who finish a script and want the quickest feedback possible.  Let’s face it: screenwriters are insecure about their work and love the idea of not having to wait long to receive feedback. But is quickest necessarily the best?  Let’s take a deeper dive.

How useful are the various sections?

Logline: For those writers who labor over creating a logline, the logline provided by this AI service was pretty good. It captured the essence of the story in one sentence. However, the analyst I use is just as good at loglines.

Genre: It was accurate, but obviously a no-brainer.

Top Keywords: No need for this at all. I don’t even know why this is included – unless the writer is planning on incorporating key words somewhere for purposes of SEO.

Location Setting: The script I submitted has numerous locations, but the report included only one. Sure, the number of locations is relevant to the budget (and producers obviously like to get a feel for that), but as a report for writers, it’s not. They should already know the accurate number of locations/settings.

Script Score: My script received an overall score of 8.5 with each category receiving high marks.  There were brief comments complimenting the execution of each area (Character Development, Plot Construction, etc.), but there was nothing there that shed any light on actionable improvements I could make. I know scoring is something writers like to see – and I’m sure producers and screenwriting contests use them to weed out which scripts to advance in their selection process. But is it actually useful to writers in terms of rewriting a script – which, to me, is the whole point of paid analysis?  More on this later.

Synopsis: It was accurate and well done. No problem there. Again, however, I’m looking for actionable information, not mere regurgitation of my plot.

Comprehensive Synopsis: Also accurate in the details, but I would expect nothing less with AI (or a human reader).

Plot Assessment and Enhancement: Okay, this should be the most important section of the report with actionable notes. However, what I found – especially when comparing to my human analyst – was that everything mentioned was very general and lacking the details I would need and want to help with a rewrite. And again, for a writer, this is the purpose of paying for analysis.

To be more specific, this section was broken down into two sections: “What’s working well” and “What could be improved.” The former section was basically a repeat of what was covered by the comments under Script Score. The latter included these comments:

“Refining some plot points and character motivations, along with tweaking the balance between humor, drama and action would help strengthen the story.”

“The pacing slows down a bit in the middle. Tightening this section could help momentum.”

“The ending could provide more closure on key character arcs.”

There really wasn’t much more than that – and those comments and a few other general ones would obviously be of no help in a rewrite.

To be fair, this brief section ended with “Please let me know if you would like me to elaborate on any part of this assessment.”  I guess I could have responded and asked them to elaborate, but I didn’t. Would my response have required them to have a human read my screenplay and elaborate? Or is the AI program able to dig deeper in response to a writer’s questions? And would it cost more money? I’m not sure – and maybe I will ask for elaboration and report back. With that said, I never really felt the need to have my human analyst elaborate on her reports because they were already extremely detailed with clear, specific and actionable information.

I should add here that my human analyst always includes a section on dialogue with detailed notes on specific lines and scenes that work and don’t work. I may or may not agree with all her notes, but I can honestly say that I always find at least a few gems in her notes that have made a big difference in my revisions. There was nothing like that in this AI report.

Character Profiling

This section went through every character with a speaking part with comments pro and con. Once again, I found the positives to be the same as already mentioned in the report. As for the Areas to Improve for each character, I also found them to be generic, similar to the Plot Assessment and Enhancement section.

Main Character Casting

In this section, which is one of the longest of all, the report provides a list of actor suggestions for each role and why these actors would be right for a given part (including lists of films the actors have starred in). Names suggested included Timothee Chalamet, Lucas Hedges, Tom Holland, Mark Ruffalo, Ethan Hawke, Paul Rudd, Jeremy Renner, Ben Affleck, Josh Brolin, Bradley Cooper, Jake Gyllenhaal, Jon Hamm, Brie Larson, Emma Stone, Anne Hathaway, Jessica Chastain, Natalie Portman, Amy Adams, Scarlett Johansson, Emily Blunt, Zoe Saldana, on and on.

To me, this list is a complete waste. Sure, anyone would want to have any of these actors appear in their film. But what’s the point and benefit to having this list in a spec screenplay analysis for writers? What could I do with this list? In my opinion, this is included as filler in the report, perhaps also to flatter a writer by showing what real actors could play the parts they wrote. The fact is that I’d much prefer to see more fleshed out, actionable notes to help me with my revisions.

Main Character Casting Limited Budget

This section, like the previous one, is one of the lengthiest in the entire report – and another total waste. Not only does it repeat most of the A-list names in the section above, but it adds a bunch of other actors who are lesser known (along with their credits), supposedly because this list is for a lower budget.

Not to beat a dead horse, but what’s the use of this section to a writer? Maybe some writers would appreciate seeing these suggestions because it makes them feel like someone is taking their screenplay seriously enough to suggest talent. But remember: it’s an AI program, not a human analyst, or more importantly, a human producer.

Comparative Film Analysis

Another long section, this one lists ten movies “that share similarities with aspects of this screenplay, along with their box office performance.”  Maybe it’s just me, but this is something else I don’t need to see, especially in lieu of the actionable story notes I’m “hiring” an analyst for in the first place. I guess if a writer is preparing a pitch deck and wants to include a list of comparative films with their box office numbers, it could be helpful. Perhaps it’s the same thing with the casting suggestion sections. But I maintain that a screenplay analysis should include only information related to the actual story and characters. Not filler sections that may stroke one’s ego at best.

Conclusion

I think by now you pretty much know my conclusion: As a writer, I would stick to a human analyst who digs deep into all aspects of a screenplay with actionable info to help with the rewrite process. Of course, a really good human analyst is the key, because I’ve seen some who really aren’t any better than the AI analyst I used.

Also, don’t be tempted by the super quick turnaround and lower cost of the AI services. Because a great human analyst is well worth the wait – and there are plenty of affordable options like the one I use.

Keep in mind that I’m speaking to writers here. I realize that many producers and screenwriting competitions already use AI analysts in order to narrow down the stacks of submission they receive. Perhaps some look at only the numerical score to determine if a screenplay advances to the next step, whether that be a human development exec or the next round of a competition. Although I believe numerical scores don’t necessarily indicate a screenplay’s potential, I do understand that some of the large contests and producers may prefer the ease of the AI process that eliminates the hassle of hiring dependable freelance readers at a low wage. For my money, however, I’d ensure a contest is using human readers before entering.

I should add that this was my single experience with one AI service on one screenplay. Maybe this was an anomaly. However, as a human and humanist, I’m always going to prefer empathy and life experience over technology in any artform.