EB.
You Send Steve Your Deck. He Sends Back ChatGPT's Thoughts On It. - EB.

You Send Steve Your Deck. He Sends Back ChatGPT's Thoughts On It.

Date: Oct 7, 2026 Author: Eytan
Reading time: 3 minutes Tags: AI Writing Workplace Content Marketing

Summarize in... 🤖 ChatGPT | 🔎 Perplexity

tldr

If I wanted AI’s opinion, I’d ask it myself. Five rules for giving and receiving feedback with colleagues now that everyone has a chatbot: write it yourself first, don’t forward AI’s take without explaining it, don’t force people to fall back on AI feedback, don’t regenerate someone’s work and call it a revision, and ask what constraints someone’s under before you hand them AI output.

You send Steve your deck. He sends back ChatGPT’s thoughts on it. With a 🙏.

First, huge congrats on not turning your old Furby into a voodoo doll of Steve. I’m proud of you.

Here are a few rules (fine, suggestions) for working with colleagues in the age of AI without being a jackass.

TLDR: if I wanted AI’s opinion, I’d ask it. It’s right there, and it doesn’t make me say “hi, how was your Sunday” before my ask.

1. It starts with you. Don’t be lazy. Write it out yourself first. AI can’t do your thinking for you. Even if it could, it shouldn’t. Be proud you’re a human meatbag. There’s a ton of research showing that writing is how you find out what you think. Open a blank doc and be bad at it for ten minutes. The cursor of doom is a learning experience.

2. Don’t forward people AI’s take on their work without explanation or a deep dive. They have AI too. Two lines of feedback with a typo beats a college essay based on those two lines. And inflating a note with AI so someone else can shrink it with AI is two robots having a lovely chat while the humans get a photocopy of a photocopy. Send the core thought.

3. Don’t force people to fall back on AI feedback. Be deliberate about what you ask people to read. Human attention is the one thing nobody has figured out how to generate. Treat it with the reverence the last slice of pizza should get. Creating is easy now, so sending half-baked work for feedback is outsourcing your job. If it took you 30 seconds to make, I’ll give it 30 seconds of feedback. Put your time in and people will put theirs in.

4. “Here’s what I meant,” with the full work regenerated by AI, makes people hate you. A lot. Without an explanation of the changes, it’s the RC Cola of feedback — right category, laughable attempt. Steve, for the love of Transformer technology, nobody has any godforsaken clue what you changed, why you added colons and em dashes in every line and a half, or why your design has more eyebrows than Frida Kahlo (honestly, I’m proud of that line).

Managers, your job is making sure people understand what’s good and what isn’t. Give ’em a critique fishing rod, not your AI-generated fish.

5. Anything you can generate, anyone can. Before you generate your own (yes, adding a panda bear to the company deck was very clever), ask the person what constraints they’re under and what guides them. Hell, see if they have input for your prompt. Marketing wants consistency, product wants the feature on screen, R&D thinks it’s all about features — and they’re all right from where they sit. The more you know about what they want, the better the work.

(And Steve, if you’re reading this, loved the deck feedback 🙏)

P.S. Some of you are Steve. Admit it.


This post started as a LinkedIn post. If you want more of this in your feed, follow me on LinkedIn.


Let's make a deal. Drop your email and I'll let you know when I post new insights.

Let's make a deal. Drop your email and I'll let you know when I post new insights.