Let me preface this by saying: I am not new to AI tools. I have paid, tested, praised, and criticized my fair share. But this? This is beyond frustrating. It is exhausting. And frankly, unacceptable.
As a ChatGPT Plus subscriber, I pay $20 each month with the clear understanding that I will receive priority access, faster response times, and more customization options. What I did not expect was utter disregard for the most basic instructions—instructions that I explicitly provided, that the system acknowledged, and that were supposed to apply across every interaction.
My expectations were not wild. Among the simplest and most direct of my instructions?
“All generated images must be in JPG format, not PNG.”
Seems easy enough, right?
And yet, image after image is returned in PNG. Over and over again. Even after confirming that these rules were to be globally applied. Even after documenting them clearly, with numbered items. Even after being reassured they would be followed. If this sounds petty to some, let me break it down:
When I request a JPG and I get a PNG, that’s not just an oversight. It’s a failure to honor customization—one of the key selling points of the paid subscription. It forces me to use third-party tools to fix something that should never have required fixing to begin with.
It’s not just about file format. It is about trust. Consistency. Respect for the user.
If ChatGPT Plus cannot follow a simple instruction to generate a JPG instead of a PNG—after being explicitly told to do so—then what is the $20/month even for? What does “Plus” mean if not better functionality, greater control, and actual responsiveness to user preferences?
I’m not paying for enhanced performance.
I’m paying to correct the AI.
I’m paying to remind it of what it already agreed to do.
I’m paying for frustration—and I get billed monthly for the privilege.
The irony? I have been using ChatGPT to streamline my creative and professional work, including crafting political commentary, legal analysis, and AI ethics content. But I now find myself using other tools—yes, free ones—to accomplish tasks ChatGPT refuses to get right, no matter how many times I say it.
So OpenAI, if you are reading this, here is my very public question:
What am I actually paying for?
Because if this is what $20/month gets me—missed instructions, inconsistent results, and radio silence on persistent bugs—I would have been better off using the free version and outsourcing the rest.
As I see it, OpenAI has two choices:
- Start taking user-defined instructions seriously and treat them as persistent, non-optional parameters
- Be transparent about the limitations of “custom instructions” so users can make informed decisions before shelling out money
Until then, I will keep asking: Is frustration a premium feature now? Because I’m getting a lot of it.
Written by: Jay Santana
Davenport, Iowa
Blogger. Advocate. Frustrated Paying Customer.
jtwb768.com
jtwb768b@gmail.com
