Of course beating injured horse, but I'd love to see ten pages of notes on "Premise."-- Your premise is good; your premise is bad; your premise has been done; your premise is unique; your premise is absurd; your premise is delightfully absurd . . . I mean, I honestly would like to see how the robot gets ten pages. if Specific Notes could do a post, or a link to a post, of a ten-page example of a premise critique, I would consider sending them a few bucks for the effort. A challenge!
-
ATTENTION! If you do not agree with the type of service that is being offered and/or are not interested in what is being offered, please do not post in the promotional thread and simply ignore it, otherwise, your post will be removed. Thank you.
website Specific Notes is open.
- Thread starter Specific Notes
- Start date
Yes, we will post an example on our website in the next few days. 
First I want to thank you for staying engaged here. I admire that.
This is exactly what I feared:
The original shows character and individual speech patterns. The revision
is emotionless and even directs the reading of the line. There are thousands
of screenplays out there that use this method. I guess that's how the AI learned.
I found the notes to be so "by the book" that they suggest removing all
the uniqueness form the dialogue:
direct lines" might be a text book note but in this case it's way off the subtext
of the character.
But the the end the AI generated note does say:
change "the script's most potent elements".
This is exactly what I feared:
The original shows character and individual speech patterns. The revision
is emotionless and even directs the reading of the line. There are thousands
of screenplays out there that use this method. I guess that's how the AI learned.
I found the notes to be so "by the book" that they suggest removing all
the uniqueness form the dialogue:
Yet, the strength and uniqueness of the dialogue IS the banter. "simpler, more
direct lines" might be a text book note but in this case it's way off the subtext
of the character.
But the the end the AI generated note does say:
Changing the interplay and vivid dialogue to that staccato revision would
change "the script's most potent elements".
The biggest limitation in AI is that it will *ALWAYS* do what you ask it to do.
If you ask for constructive criticism about dialogue - you're gonnna get a critique!!
AI will never, ever, ever, respond with "Woah this is a masterpiece! Such witty dialogue, I wouldn't change a word!"
Due to this limitation, feeding high quality scripts into an AI results in criticism without integrity.
If you ask for constructive criticism about dialogue - you're gonnna get a critique!!
AI will never, ever, ever, respond with "Woah this is a masterpiece! Such witty dialogue, I wouldn't change a word!"
Due to this limitation, feeding high quality scripts into an AI results in criticism without integrity.
I agree; me too. I don't have time, right now, to look at the example, but will try to critique the critique this evening. Thanks for the response (and it was so quick, i wonder . . . Are you, Specific Notes, a robot?
) anyway.
Last edited:
Which is why it is essential to have access to the interface being used so you can reword, ask differently, and feed new changes it gave you back in, etc. This is a "one-off" service where they utilize such AIs and you pay them for feedback once, after they themselves do the rewording, human edits, etc.
What sort of query are you supposing, where an LLM is capable of assessing if dialogue is truly as good as it can get, or not?
I'm not veering anywhere, I'm standing my ground and saying that The biggest limitation in AI is that it will *ALWAYS* do what you ask it to do.
Is the answer the same? Not the point. The point is that if you ask for a critique, you will always get a critique, whatever that critique may be, however many different critiques you recieve, it will always be a critique!! Even if you give it casablanca its gonna give you a critique and tell you to change the dialogue. Something to keep in mind with all AI services, this one included.
Edit to say -
This is actually a minor limitation - In other words I'm saying this critique isn't valuable if you have a perfect script like casablanca.
I mean no shit, right? no feedback is much value if you already have a perfect script. Still good to keep in mind how these tools work if you're gonna use them and to understand their limitations. I bet if he used an example with a worse script, he would have a better example result.
Last edited:
Forget the major screenplays not talking about that. If YOU have access to the AI you can ask to critique how you want. And you can try many times different ways. Using the word slang, or dialogue, or conversational. That's all. If you are paying a service to run it though AI you get one report back based on the way they word the query. Simply put, you are paying a human to do what you could do yourself multiple times with a subscription. Is that AI? Or a labor cost at a premium and calling it AI? They obviously do not have their own AI or they would have an interface to test. It's a reselling business.
Yeah reselling, middleman, exactly. Biggest thing you're paying for is the convenience.
Typical chatGPT there is a character limit, you can't copy/paste an entire screenplay into the dialogue box.
There are ways around that, tokens, API, installing local, etc, but it's less convenient.
Ehhhh ... I'm not sure it's a great idea to choose Casablanca as an example of the "perfect" script!
On the contrary, if an algorithm had been let loose on the mess that was the original (and every intermediate) Casablanca script, it would probably have ended up as just another forgotten movie from the '40s.
OK, that looked at first like a positive response ... but it's of no real value at all if it doesn't show a side-by-side comparison of the original and any/all proposed revisions. At a quick scan, the analysis contains about 20 references to specific lines or passages of dialogue in a script of unknown length (but at least 93 pages!) and is overwhelmingly made up of boiler-plate statements of the kind found in any guide to creative writing.Yes, we will post an example on our website in the next few days.
Whether or not they are relevant to the dialogue in this script is impossible to say or discuss as almost every suggestion is far to vague to be of use. Either the example is so precise, one cannot know how many times it occurs throughout the script; or it's immediately countered by a "However ..." in the next paragraph. If that's an accurate example of the "laser focused" approach promised, one is still left with the job of re-reading and correcting one's own script ... which isn't particularly focused at all.
With regard to the one specific example of original and corrected dialogue, I'd also second @directorik 's comments. The analysis initially celebrates how
yet the one and only worked-through revision turns this:
You used to be like, different, you know? You used to be so emotional. You used to cry about everything.
into:
You used to be different. Emotional. You cried at everything.
Seems like sacrificing the messy, human, repetitive nature of "how people actually talk" for the sake of a "rigorous approach to punctuation and formatting". Ignore everything said up to this point about crackling wit and vividly drawn personalities ...
OK, that looked at first like a positive response ... but it's of no real value at all if it doesn't show a side-by-side comparison of the original and any/all proposed revisions. At a quick scan, the analysis contains about 20 references to specific lines or passages of dialogue in a script of unknown length (but at least 93 pages!) and is overwhelmingly made up of boiler-plate statements of the kind found in any guide to creative writing.
Whether or not they are relevant to the dialogue in this script is impossible to say or discuss as almost every suggestion is far to vague to be of use. Either the example is so precise, one cannot know how many times it occurs throughout the script; or it's immediately countered by a "However ..." in the next paragraph. If that's an accurate example of the "laser focused" approach promised, one is still left with the job of re-reading and correcting one's own script ... which isn't particularly focused at all.
With regard to the one specific example of original and corrected dialogue, I'd also second @directorik 's comments. The analysis initially celebrates how
yet the one and only worked-through revision turns this:
You used to be like, different, you know? You used to be so emotional. You used to cry about everything.
into:
You used to be different. Emotional. You cried at everything.
Seems like sacrificing the messy, human, repetitive nature of "how people actually talk" for the sake of a "rigorous approach to punctuation and formatting". Ignore everything said up to this point about crackling wit and vividly drawn personalities ...
You and Rik both noticed the same thing.
The AI compliments the well written dialogue, but then suggests changing the very dialogue it complemented. Why??
Now you have seen and identified exactly what I was talking about.
The fundamental weakness of AI - you asked for a critique with dialogue changing examples - so it gave you one.
It will NEVER say
p.s. -- compliments is spelled with an i - but complemented spelled with an E - what kinda english fuckery is this??
Complimented is past tense of compliment.
Okkkay I see what happened now.
I spell checked myself with google - if you google "complimented" it says Did you mean complemented
I didn't look any further than that, I just assumed I had spelled the word wrong since google corrected my search query.
Turns out, no, I did not mean complemented.