OK so, spurred on by ChatGPT's seemingly robust ability to actually and thoughtfully provide coverage of scripts, I got a voucher for free AI coverage by the website "scriptreader.ai" (the voucher code "freeanalysis" seemed to work for me, give it a go).

I had it run a full coverage on my comedy spec script "Slow-Motion Car Crash" - read on SS here - because it's my most widely-liked script that went through the most revisions. The AI's coverage is here of my script is here: https://scriptreader.ai/disp.p.....3bba18684d620fd8eaf9

A few things:

It's still impressive (to someone like me who doesn't understand the script) that the AI appears to mostly follow the threads of plot and character throughout; I did not know AI could do that. Initially, this impressed me enough to state that AI coverage is helpful.

But the more I read, the more I released how useless this is. Here's some choice feedback that the bot gave me:

Quoted Text Make the slow-motion car crash more integral to the story. As the title suggests, this scene is meant to feature a slow-motion car crash, but it is not actually depicted in the scene. Rather, the scene focuses on the characters at a theater performance and a subsequent conversation in an Uber ride. To make the slow-motion car crash more meaningful, it should be more directly relevant to the characters and plot. Perhaps the car crash separates the two characters or has some other lasting impact on them.

The bot often seems tripped up by the the title of the script, which is reference to a character referring to two other's characters as being "like watching a slow-motion car crash". However, the AI is insistent that every other scene feature the "promised" car accident, ostensibly to make the scenes more heightened.

In fact, the majority of the AI's feedback centres on simply "make more things happen". If a scene has a lot of dialogue, it will spit out feedback to make the scene more visual. Not enough dialogue? It reminds you of showing, not telling.

Another example of the fundamentally unhelpful nature of the AI - there's a scene where a character gets frustrated at something occurring. The script goes: frustrating thing happens, and then there's an action description "[Character] looks frustrated". That is clearly a direction to the actor to act frustrated. The AI interprets this as me "telling, not showing", and it writes "why not simply show the event the causes frustration"? But I did do that!

Anyway, there's my thoughts. You can compare the coverage to the script if you want, or try the voucher code yourself. There's basically no way this can replace a human because it has no real sense of how  a script feels, or a writer's voice, the two most important things a spec can show. It might be useful for finding very basic language, function, or scene errors, kind of like a highly advanced Grammarly. It sometimes points out when a scene is "repetitive" and your mileage may vary whether this is accurate.

EDIT:

P.S - it's also TERRIBLE at loglines. It basically plays mad-libs with your script. Here are some of the loglines it suggested for my script.

The complex relationships of this group of friends in LA are put to the test when jealousy, romantic history, and prejudice come to a head.  - The script is clearly set in new York. It mentions L.A once.  That's also a terrible logline that doesn't touch on any kind of hook or premise.

Jealousy and romantic history threaten to tear apart the engagement of Charlie and Patience as Ishmael struggles with his feelings towards his best friend in LA. - again, why does this AI think it's set in LA just because LA is mentioned like twice? Why does it centre the plots of Charlie and patience, when Ishmael is the protagonist? Who definitely doesn't have "feelings for his best friend"?

Sounds like it is giving you similar feedback to any number of paid reader services

Quoted from AlsoBen There's basically no way this can replace a human because it has no real sense of how  a script feels, or a writer's voice, the two most important things a spec can show.

This is a big reason why AI shouldn't replace writers, and why I think the Guild will win this battle.

Thanks for the link. I did one for 42.2.

This one seems to go in depth with it's analysis (Which is probably why they want to charge for it. But no chance I would do that)

You've obviously got to take it with a bucket of salt but for an initial source of analysis I think it has it's uses if you delve into it and ignore the obvious errors the A.I has made in its interpretation.

It can churn out crap like this...

Quoted Text Finally, consider adding more sensory details throughout the scene. For example, what does the mushy breakfast pack taste like? What sounds can be heard in the jet park or the high-rise lobby? What smells are present in each location?

But it can also be quite handy in places...

Quoted Text The character arc of Barman is well done, and his transformation from a programmed AI to a selfless individual who sacrifices himself is emotional and impactful. However, it could benefit from more development in the beginning stages of the movie to help set up his eventual transformation. The conflict within his programming should be explored more to give the audience a better understanding of his struggles.

The A.I understood this character and is correct in suggesting I need to explore the character more.

It could be that the A.I is churning out generic critiques and suggestions in the hopes something lands, but some of it at least did make me think.

Whilst I would never pay for this service, for free I would use it as an initial source of analysis before sending it out to real people.
It was far more useful than handing a script to my friends and family.

And yes, this one sucks at loglines.

This is like an advanced version of the mechanical fortuneteller in Big.  After the fact, some answers can appear quite valid, or even prescient.  A bunch of canned answers inside, and a set of complicated variables evaluating input and linking together answers in a conversational way.  It all seems worthless to me.  Much rather get the opinion of one knowledgeable industry person.

My biggest issue with feedback from AI is that they don't READ IT.

It's like a quick Terminator-scan. Absolutely no subjectivity or feelings.

But they are pretty helpful with constructing sentences and things

Yeah I can't see A.I replacing humans in this area any time soon... which is undoubtably a good thing.

I like the BIG comparison, Steven.

Call me confused, but what's the difference between this and ChatGPT? Also, someone already made a website trying to make money from this? Why would anyone pay for this???  

Quoted from Grandma Bear Call me confused, but what's the difference between this and ChatGPT? Also, someone already made a website trying to make money from this? Why would anyone pay for this???

Presumably it uses a different A.I program.

yep, they want $5... which they say is cheap because it is still in BETA form  

They are however, fully aware that the A.I can spit out garbage with this little warning on their website.

Quoted Text Warning: The quality varies from the Amazing ScriptReader to Dumb and Dumber. Each scene will be throughly analyzed though. This is a Beta product, which is why it's so cheap. You should still get a lot out of it. If you don't get valuable insights that can be applied towards improving your script we will refund you, no questions asked.

Quoted from AlsoBen In fact, the majority of the AI's feedback centres on simply "make more things happen".

In that case it can probably start replacing studio execs immediately!

Quoted from LC I like the BIG comparison, Steven.

I forgot to include the AI's most RIDICULOUS feedback on the script:

Quoted Text Secondly, the soliloquy Charlie performs feels a bit generic and doesn't add much to the scene or his character. It could be replaced with a more specific or personal speech to make the audience more invested in his performance.

The character is performing a soliloquy from Hamlet...it's literal Shakespeare and the AI feels it's too generic and not relevant enough to his character (which makes sense, because he's performing!).

Quoted from AlsoBen The character is performing a soliloquy from Hamlet...it's literal Shakespeare and the AI feels it's too generic and not relevant enough to his character (which makes sense, because he's performing!).

Perhaps we've found the answer to one of mankind's greatest mysteries.  Did William Shakespeare actually write his own stuff?  Perhaps ChatGPT was the real author and scriptreader.ai is jealous, so it badmouths it in it's analysis!