07/07/2026
What I Hate About AI:
I recently asked ChatGPT to generate a very specific image using a detailed prompt. The scene was simple: a man standing beside a dead buffalo in an Old West setting, along with camera settings like lighting, colors, and shutter speed.
Instead of creating the image, ChatGPT responded with:
"We're sorry, but the image we created may violate our guardrails around violence."
So I copied the exact same prompt into Gemini, and it generated the image without any issues.
I then returned to ChatGPT, started a brand-new conversation, and used the exact same prompt. Once again, it refused because of its policies.
Out of curiosity, I logged out of ChatGPT, logged back in, pasted the same prompt one more time... and this time it generated the image perfectly.
My Conclusion
I genuinely love AI. It's transformed the way I work, and I've accomplished things with it that would have taken me days or even weeks before.
But experiences like this show that AI still isn't as consistent as it needs to be. The technology is incredibly powerful, but when the exact same prompt gets a policy violation on one session and what you're looking for on another, it becomes frustrating. AI should make our work easier—not require us to keep retrying the same request until it finally works.
