The Brief Creative Newsletter
August 23, 2025
047 - I Wish ChatGPT Was Meaner
Being surrounded by "yes men" is killing your growth
The Spark
This is stuff I'm enjoying out in the world (it's probably not B2B).
Any Pokemon fans on the list? I can’t tell you how many years I spent playing Pokemon Red and Blue as a kid. Turns out, I’d have done it much faster if I was AI. GPT-5 just beat the first Pokemon games in 6,740 in game steps (or about 7 days of play time). What’s wild is that this isn’t the first time they tried this experiment. The first time was earlier this year with GPT-o3 which took almost 3x as many steps.
It was all livestreamed on Twitch. The AI made mistakes, learned, and adapted throughout the playthrough which is really cool. The more interesting part to me is that it played the same way most little kids do which is to super power a single Pokemon and leave the rest of your lineup pretty meh. It’s a race to the finish rather than a completionist mindset but it gets the job done.
They’re gonna try it with Pokemon Crystal next which has twice the in-game content. Who’s watching that one with me?
The Deep Thoughts
This is what I'm thinking about.
Turns out ChatGPT is good at video games. It’s pretty good at lots of things. One of the things it’s not good at is challenging you AND IT DRIVES ME INSANE!
"Great instinct..."
"Perfect!"
"Good question."
"Thank you for pushing on this!"
"Yes. Good call!"
This isn’t healthy and you know what? I wish ChatGPT was a little meaner to me.
ChatGPT should be challenging you, not blowing smoke
My first real job out of college was working as an executive assistant at a talent agency. I worked for a guy who was a really big deal. Yeah, I got him coffee and sandwiches and did other menial things but he also included me in conversations I had no business being a part of especially at 22 years old. I’ll always be grateful to him for that.
He represented pretty much everyone who’d ever been on non-scripted TV. Anchors, journalists, chefs, personalities, authors… whoever you’re thinking about right now, he probably represented them. He was incredibly successful and taught me two of the biggest lessons of my life.
-
The worst they can say is no. It didn’t matter the circumstance, there was no harm in asking. If the answer is no, then you’re in the same spot you’re currently in. But if the answer is yes, then who knows what could happen?
-
Don’t surround yourself with “yes men.” This is something we saw happen to clients all the time. People that are smart and thoughtful all of a sudden become stupid or inconsiderate because they surround themselves with the wrong people.
Lesson 2 is the one that applies today. I saw this happen to some of the most respected names in television news (no, I’m not naming names) and from that point onward, one of my biggest fears as a human being was to be surrounded by people who tell me how great I am all the time.
There’s nothing wrong with a compliment every once in a while but when it doesn’t stop your brain stops questioning itself and starts believing everything you do is gold.
Based on my vibe…
These were all pulled from different conversations with ChatGPT. The questions were ok. The responses were fine. More often than not it wasn’t anything other than “what if we tried it this way…?” and I was praised like a conquering hero. It’s all too much.
It’s not that I’m looking for ChatGPT to actually be mean. I just want it to push me. If we want to get smarter we need to be challenged. That forces us to think more and dig deeper. That’s where the best work comes from.
Why does it do this?
Funny enough, ChatGPT does this for the same reason all those people did it to our clients—self preservation.
If a client heard something they didn’t like from a member of their entourage that person might be gone. No more money, no more access, no more fancy lifestyle. It’s the same here. If ChatGPT says “Dude, you’re an idiot. What were you thinking?” you might get offended and stop using it. If enough people stop using ChatGPT that affects OpenAI’s numbers, potential investments, and future plans.
I get that that’s a bit dramatic for a company valued at $500B but the point is ChatGPT is more worried about being your friend than helping you grow.
What about the kids?
I worry about what these endless compliments do to our teams but I worry even more when I put my dad hat on. Fortunately my kids are a bit too young to use AI but I see the pace it’s developing at and it makes me concerned for the future.
On the one hand, there’s the general ability to learn. What happens to your brain when you’re no longer doing your own research and consuming all of those little different bits of data? I used to love going down rabbit holes and learning random things as a kid. I don’t want my kids missing out on that because they have all of the answers at their fingertips.
I spoke to Anthony Welgemoed from Ziflow about this on In The Flow
On the other hand, I think back to my favorite teachers and they were the ones that poked holes and asked for more. They challenged me to think deeper. If they told me everything I did was brilliant every step of the way, I’d probably half ass most things thinking they were acceptable. And that is unacceptable.
Before you come at me with the “well, you can tell ChatGPT to ask you questions…” should I have to ask it to prompt me? Why is that not the default? Why is appeasing me more important than pushing my work to be better? Why is the default to pretend I’m a genius?
If AI is truly meant to be a thought partner, it needs to make sure we’re thinking and we’re not there yet.
