top of page

"hey chatgpt, u up?" The three Fs that absolve my AI writing guilt

  • Writer: Paige Schwartz
    Paige Schwartz
  • Oct 22
  • 7 min read

As the CEO of a B2B tech writing agency, my relationship with AI is best summed up as “it’s complicated.” 

After swearing I’d never get involved, I have to admit, the attraction is strong. I can’t dismiss AI as “slop” or even “never as good as a human writer,” because frankly, it can write some pretty excellent prose (I don’t kiss and tell, but some of my best-received work lately has been highly LLM-assisted). On the other hand, there are situations where I wished I never wrote ChatGPT that “u up?” text. 

Now that we’re past the fling stage, I’ve established some boundaries in my relationship with AI. Now I feel confident and secure I’m spending my time and energy wisely when AI-assisted writing meets these three criteria: 

Fun: Must be as fun or more fun than writing it from scratch

Fire: Must impress me, and must impress others

Fast: Must be faster than doing it myself 

These principles might help you too (that’s why I’m sharing them!), but let’s be clear, they’re far from perfect. It’s complicated, remember? These guidelines dodge deeper philosophical or ethical concerns, and they’re based almost entirely on personal experience. They simply feel right for me, in this moment — and they’ve made it so nothing I pair-write with AI is a doc of shame. 


Fun

If you’re not having fun, you are giving up something about the craft that you love.

Like most (perhaps all?) professional writers, I’ve made this my career because it feels like play. I was afraid that AI would steal that joy from me forever, and turn me into little more than a human-in-the-loop. I let that fear make me a skeptic — and to be fair, there have been many reasons to be skeptical, not least the fact that the writing quality just wasn’t very good, even until recently. But then I tackled some major AI-assisted projects this summer — and I found that pair-writing with AI can be challenging, surprising, enlightening, and engaging. It’s actually fun.

If you’ve felt this spark, you get it. The moment when something clicks and the LLM writes exactly what you had in mind — it’s undeniably thrilling. As you form a kind of understanding of how its “brain” works, sometimes its quirks and off-beat suggestions can make you laugh out loud. Getting what you want out of it is ultimately (for me) a fun, zero-stakes exercise in good communication, one that’s translated to other aspects of my work and life. And for fellow nerds, it’s just fun to be learning a new tool (though tool doesn’t feel like the right word — it’s more like learning to collaborate with an alien). 

Working with AI isn’t always fun. If we aren’t clicking, I restart the project on my own. But the fun principle has helped me quiet a lot of anxieties about using AI. Intrusive thoughts like, am I giving up something that made me “human”? Am I training my own eventual replacement? I just think: Well, this is my life, and who knows what’s in store, but right now, I’m having fun. 


Fire

If it’s not amazing, it’s self-sabotage. (I know “fire” is pretty cringey coming from a millennial mom. I’m sorry, I needed an F.) 

Here’s an idea I’ve found freeing: When people pay you to write they aren’t paying you to literally put down one word after the other. They’re paying you to think about a thing and produce a written document that reflects your judgment. As the editor-in-chief of Copytree, the one thing that has gotten me this far (and that I truly trust) is not my ability to write, but my ability to tell what is good. If it’s fire then it’s fire, if it’s not then it’s not. 

And you can bet I apply this to anything that I’m pair-writing with AI. I am ruthless with the red pen and I may completely rewrite the piece so that very little of the original remains. Far from believing AI is dulling my critical thinking skills, I feel it’s sharpening them, because I’m extra critical with everything AI writes. 

To give AI credit though, if you steer it well enough, it can balance surgical logic with flashes of genuine eloquence. Purely anecdotally, GPT-5 (my LLM of choice — your mileage may vary) is a major step forward. It parrots tone convincingly, picks up on the nuances of an argument, and is more detail-oriented and coachable than previous versions.

Now, the most important thing is that you always need a second set of eyes when you use AI as a writing tool. It’s very easy to be deluded by the quality of what you produce with AI, for the same reason as you tend to think whatever you wrote yourself is pretty dang amazing. Since you’re co-writing the piece with AI, you can’t also be the editor.


Fast

If it’s not faster than writing it yourself, you are losing money vs. using your brain to do something you like and are good at.

If you’ve conquered fun and fire, the decision to work with AI or not gets down to brass tacks: is it actually more efficient for you (and your client)?

The equation isn’t easy — it varies project to project and person to person. For the kind of work we do at Copytree, I’m probably a 90th-percentile efficient writer. And I’d rank myself as maybe a 60th-percentile efficient AI user. In the best case, and this is a complete guesstimate, it’s probably only 30–50% faster for me to pair-write with AI vs. writing something from scratch myself. 

I’m learning and getting better at leveraging AI, but I would bet that “fast” will be the hardest criterion to hit for a while still. You have to really work to get AI on your level — it’s a great listener but a terrible mindreader. Then there are the logistics: the waiting for responses, the tendency for AI to cling to a random comment you made back at the start of the conversation, the lack of great UX affordances for the pair-writing-with-AI loop. And there’s always the time you spend finessing by hand after you’ve gotten as far as you can get with AI. 

Certain pieces have taken me twice as long to write with AI than if I’d just focused and banged it out. Sometimes with ChatGPT, you look up and it’s been an hour and you haven’t made any real progress. The thing is, when you’re a writer, every hour you spend thinking about the material is an hour closer to a breakthrough. But it’s different when you spend an hour trying to help someone else write something, and they just aren’t getting it. I’ve learned the hard way: You have to cut your losses with AI early.  


Before applying the three F’s, use this question

Now, I didn’t write this post with AI. I didn’t even think about whether it would be fun, fire, or fast — I knew that I had to write it myself. How? 

It was almost subconscious until I formalized it here, but I have a question that short-cuts through the criteria for me, and it’s this: Am I writing to discover what I think and what the piece needs to be? Or do I already know what I think and what this needs to say?

If you’re writing to explore, then AI almost never works. In my experience, you can make great use of AI when you take the role of the piece’s director, coaching AI to deliver its lines in precisely the way that serves your vision (and spending quite a bit of time in post-production). But if you don’t have that confidence and that vision, the LLM is going to run you in circles finding 100 different ways to say nothing worth saying. 


Don’t ask, don’t tell

These principles have brought some much-needed definition and balance to my relationship with AI. I truly feel that if I’m using AI in a way that’s fun, fire, and fast, I’m doing right by my clients (provided they haven’t asked us to avoid AI!), my team, and myself. 

Unfortunately, the perception in society at large, and in many vocal corners of the writing community, is that using AI to write is cheap and inauthentic. And some people have real moral concerns with AI (resource usage, IP infringement) that I’m not going to wrestle with here. But often, I feel the negativity comes from something less considered — from a lack of understanding of what it really takes to write something well with AI. It’s like many people today have two ideas of a writer:

  • Person toiling away at the blank page, finally wringing pure, original thought from their brain

  • Person plugging in a prompt to AI, sitting back and sipping a cup of coffee, then copy-pasting it into their Google Doc to send to the client

This misses the entire spectrum of how writing actually happens with AI (and the fact that great writing has never happened in a vacuum). It leaves out the critical thinking that makes something fire, and the creative collaboration that makes it fun. And of course it conjures up shame and fear for writers and the people who employ them — no one wants to be thought of as lazy. 

As a result, there’s a bit of “don’t ask, don’t tell” going on in our industry. Heck, I don’t even want to share the links to the more AI-assisted pieces of writing I’ve done lately, because I believe that the knowledge that something is written with AI makes you more likely to judge it negatively, even if it’s fire. 

I think the culture will change, as more people share how they co-create with AI. Meanwhile, let’s give ourselves some grace — it’s complicated. We’re all just trying to figure it out.

©2025 Copytree.

Website made by writers, not designers ;)

bottom of page