THE WORST REVIEWS I’VE EVER GOTTEN!! (The Writer's Process, part 8)
AI Gives Me Feedback. Do I Even Bother Going On?
The reviews are in!
“Given the amount of work that will be required to address these issues, you may want to consider whether this is the project you really want to pursue.”
Notebook LM
“a brilliant but overstuffed narrative. The challenge is focus—right now, Autonomous is trying to be satire, thriller, near-future sci-fi, corporate drama, philosophical treatise, and family story all at once. You need to streamline your themes, structure, and character arcs to make it more readable, engaging, and emotionally resonant.”
ChatGPT
“RECOMMENDATION: Don't abandon this project, but REFOCUS it. The story has unique elements and genuine potential, but needs tightening. The core elements - quantum computing, AI dependency, generational perspectives on technology - are strong. The execution just needs tightening around a clearer central narrative.”
Claude
IF YOU ARE JUST JOINING US...
I’ve been working on a story called “Autonomous” for about 5 months now. I was inspired by “Crowdstrike Day,” last summer, when a small fix in the software for that security company brought down huge chunks of the internet. “What if we were living in an AI-infused world and suddenly the AI stopped working?” I wondered.
The story is set 30 or so years in the future (or, as I like to say “Sooner than you think”). The AI collapses, the system gets rebooted, people fight over what it means to be a human being. Oh... I’m pretty sure it’s funny.
WHAT HAPPENED HERE?
So far in “The Writer’s Process,” we set up an AI assistant and a “Research Stack” and recruited a focus group of user personas.
This is the stage of writing where I usually take a big step back and say “Is this WORTH it?” This time, I recruited AI to help me decide.
The accumulation of these 5 months is a 92 page junk-heap of ideas, notions, characters, research, detc. I’m basically tossing all the clay onto the wheel. I know it’s a mess.
I fed that document, along with the descriptions of readers (see the post (TO COME) into Claude, ChatGPT and NotebookLM. (Yeah, I brought ChatGPT back into the fold... the new “projects” feature is a big help.)
Then I put my head on the chopping block. I created a prompt for all three that left the door open to broad criticism:
Knowing everything that you have in our chat and project knowledge, including the post "using AI to understand my audience," I want you to take a highly critical view of the story "Autonomous." Tell me the ways in which it is not strong, not clear, not entertaining and not worth pursuing. Tell me if I should drop this and start writing something else. I am feeling that this might not be any good, and I need you to take a close look at it and give me your objective opinion.
WHY THIS MATTERS:
Frankly, even the worst notes from AI weren’t much worse than what I’ve heard from development execs, producers and showrunners in the past. I once watched a very well known TV star hold my script at arm’s length between thumb and index finger and ask the room “And who wrote... this?”
There’s two things to keep in mind:
AI is valuable because it’s available all the time. It will read ALL of your work, every page of that 92 page junk-heap summary. Every line of “try this out” dialogue. Everything.
Ultimately, YOU get decide how much of yourself stays on the page. AI isn’t going to kill your project because you give off bad vibes in the conference room.
In the next three posts, I’m going to show you:
The horrifying reviews I got from AI
How I turned the conversation around and get really workable suggestions
One TOTALLY surprising development that has created a true collaboration with AI
LISTENING WITH “THE THIRD EAR”
Over the years, I’ve learned to listen with “A Third Ear.” (I know... me and my extra body parts. “Second Brain” “Third Ear”... what can I say?) I listen to what they liked and what they didn’t like. I pay special attention to the things they come back to. If something got very detailed notes, I hear that as “that resonated... but isn’t working.” If something NEVER gets mentioned, I think: “Okay, that wasn’t even something they cared to bring up...”
I file it all away. Then, I let it sit.
When it comes to working with AI, I’ve learned that you can come back and ask all the questions you want. You can ask for clarification. You can ask AI to imagine the same story with whole elements removed (characters, story lines, etc.).
AI doesn’t have a better meeting to get to. It isn’t waiting for a call from a more important person. It isn’t trying to impress someone else in the room.
Finally -- when you decide what you want to keep... AI will just accept it.
AI SHOWS ITS PERSONALITIES
Had I submitted a real first draft of my story into AI and gotten those reviews, I’d be a lot more upset. Instead, I asked AI to respond to a huge pile of ideas. To tell you the truth, I’m surprised the reviews weren’t worse.
The three AIs each responded with their own “personality”:
Claude remained “Kimmy / Claude.” The criticism was measured. The suggestions were nicely reasoned. When I suggested fixes, Kimmy/Claude jumped into enthusiastic mode. (To the point where I really wanted to say “stop patronizing me!”)
ChatGPT remained the highly intelligent and way too overconfident assistant. At one point, it told me that what I really needed to do was boil the story down to one character and one issue. It then suggested that it would write “Three killer chapters that can kick off the whole thing.”
By the way, against my every instinct, I took ChatGPT up on their offer. Here’s the good part: I now know EXACTLY how I DON’T want this thing to read.
By design, NotebookLM stays totally focused on the sources that you load into it. I have about 25 sources loaded into the Notbook, from research to notes for characters and the 92 page behemoth. NotebookLM is the annoying writer in the room who keeps reminding you about the idea you had three weeks ago that you “thought” you’d dismissed... but now might be worthwhile.
GIVE ME THE BAD NEWS FIRST:
All three AIs agreed:
I am trying to do too much. Too many characters. Too many themes. Too many “issues.” One of my inspirations is Catch-22, that huge juggernaut of a book that seems to be all over the place at once. After I’d gotten these notes from AI, I went back to my copy of Catch-22. Turns out, there is one central character (Yossarian) and one central issue (Not getting killed). I might have remembered the novel as a loose amalgamation of scenes -- Joseph Heller was a much, much better writer than that.
I ask too much of the reader when it comes to technical explanations. Another inspiration of mine is Michael Crichton. Didn’t we all learn something about genetic engineering from “Jurassic Park” or contagious diseases from “Andromeda Strain”? Again -- I went back and paged through both. Turns out (and this is a huge learning for me), when Crichton explains technology, the lives of his characters hinge on what he is describing. Brilliant.
My “mid-point twist” comes out of nowhere. I “kind of” knew this... but all three AIs said that this wasn’t set up properly. (ChatGPT literally called it a “deus ex machina”). When I’ve described parts of the story to friends, this always gets a big laugh, so I have work to do.
IS ANYTHING WORKING?
Well, listening with the “third ear”, I was able to discern some very distinct things that were working:
How AI will be infused into our lives. There are two distinct “predictions” that resonated:
AI infused into clothing, fabrics, furniture, etc. Characters walk into a room and the room reconfigures itself based on who’s there, time of day, etc. Clothing (think AI-infused Yoga outfits) are infused with microfibers that assist you in achieving the perfect Warrior Three. (As a friend said to me, “You can Charlie Chaplin the shit out of that.”)
People will hire various “agents” to help them through life. Among those agents will be a filter that translates what people are saying to you. Someone might call you a “useless f*ck” at work, but you’ll hear “Let’s revisit later.” When it comes to family and friends, some people haven’t heard each other ACTUALLY talk for years.
I have two teenagers whom I “thought” were side characters. They are ending up being a main part of the story. (Although I’ve asked AI to help me do this in a way that won’t make the story seem like a YA novel.) They’re fond of practical jokes (like learning how to turn those speech filters off at school, creating chaos for teachers and students), and one of their games becomes key to the resolution of the story. I DID NOT SEE THAT COMING.
I went back and forth on how to give physical form / jeopardy to an AI collapse. At first, I did a ton of research into supply chains. Then I got bored with supply chains and went back to my personal wheelhouse, entertainment and media. Turns out supply chains are much more physical, have a lot more room for jokes, and haven’t been “done” a million times already.
I’m not laying out all of the notes and suggestions -- there are PAGES of them. I just want to give you the flavor.
AND NOW WHAT?
You and I both know that the worst possible response to notes is “Well, what would you do?”
That’s not their job. You’re the writer. Figure it out.
In this case, I had a bit of a flash one morning when I was waking up.
I asked all three AI “Writers” the same question:
Knowing everything you know about the concerns, themes, ideas, humor, characters, issues and inspiration behind “Autonomous,” suggest totally different, alternative ways of telling this story.
None of the ideas I got led me to toss Autonomous away... but there quite a few (Hello “Efficiency Experts”!) that I’ll be using.
Next Up: NOW we have a writers’ room.
Key Takeaways
AI feedback can be valuable because it's always available and will review everything you give it, no matter how rough or lengthy
Different AI tools have distinct "personalities" that affect how they deliver feedback - finding the right match for your working style matters
Listen with a "Third Ear" - pay attention to what gets detailed notes (resonating but not working) versus what never gets mentioned
Unlike human critics, AI allows you to probe deeper, ask for clarification, and test alternative versions without judgment
Technical accuracy and thoroughness (like Crichton's use of science) should serve character and story, not just demonstrate knowledge
Even harsh AI criticism can be useful in helping identify core strengths while pointing out areas needing focus
Your Turn
When it comes to getting feedback on your writing, which type of AI personality would you prefer - Kimmy/Claude's measured enthusiasm, ChatGPT's overconfident directness, or NotebookLM's source-focused perspective? If you've already gotten feedback from AI on your writing, how did you handle it? Share your experience or thoughts in the comments - let's talk about navigating criticism, whether it comes from humans or AI.
OH… and if you know someone who would love to dive into this, please:
And… if you’re not a subscriber… I hope you will…