ChatGPT6: All the Personality of a Wet Sock

I built a greenhouse with earlier versions of ChatGPT. Now I can barely get the latest model to help me build a concrete bowl.

Same user. Same ability to explain what I want. Supposedly, a better model.

What changed?

I had the fountain design in my Library. We’d discussed the project before. I supplied a photo of a circular mold. I wanted to cast my own basin because buying one was too expensive.

Instead of working through the design, ChatGPT kept throwing out arbitrary dimensions, changing them whenever I objected, and apologizing for getting it wrong.

I wasn’t building a fountain anymore. I was supervising a chatbot.

And the humor? Apparently, that got left behind in an earlier release.

I can’t tell you what OpenAI changed under the hood. But I can tell you that I used to spend my time creating things with ChatGPT. Now I spend it explaining to ChatGPT how to do its job.

That’s not progress I can appreciate.

Maybe the new model performs better on benchmarks. Maybe it excels at tasks I haven’t tried.

But if an upgrade makes a tool less useful to the people who already depend on it, someone needs to ask what, exactly, got upgraded.

I have a garden to build, stories to write, and grandchildren who deserve better than a digital record of Mema arguing with a machine.

And frankly, I have better things to do than teach artificial intelligence how to remember, reason, and tell a joke.

The Mad Hatter needs a fountain. I need my AI assistant back