I've been using ChatGPT lately, and honestly, it feels like it's become pretty unreliable. For instance, I've noticed that both ChatGPT o3 and o4 are hallucinating like crazy, often mentioning UI elements and features that just don't exist. It's frustrating because I had to keep clarifying that those features aren't real after asking for help with specific tasks.
Just the other day, I tried to create a shortcut on my iPhone using ChatGPT's guidance, and it kept inventing non-existent buttons and options in the Shortcuts app, leading me nowhere. I ended up giving up because the responses were just irrelevant. Similarly, when I asked about exporting a configuration from my AdGuard Home setup, ChatGPT made up a button that isn't even in the interface, which meant I had to clarify again before I finally got a correct answer.
I even tried using Gemini 2.5 Pro with the same questions, and it provided correct and quick responses without any issues. It's started to feel like it's just taking longer to get information from ChatGPT now, and after so many frustrating experiences, I've decided to switch to Gemini. I'm curious if anyone else is experiencing these problems or if I'm the only one who's hit a wall with ChatGPT?
2 Answers
I've had a pretty good experience with ChatGPT, actually. I use it regularly for various tasks, and it seems to do okay for me. Maybe it’s just about the context or the specific prompts you're giving it?
I had a similar issue! I was working with it on music lyrics, and out of nowhere, it made up lines that weren't in the song. I corrected it, and it just played it off with a 'you caught me' moment, which was kind of funny but also frustrating.
Haha, been there! Sometimes it just can’t keep track and it gets really confusing.