I use AI a lot to simplify and automate my life. I pay for and use both Claude and ChatGPT regularly. This GPT-5.4 release might just be enough to get me to go all-in with Claude.
OpenAI apparently learned a lot about clickbait in this last release. Nearly every response ends with a “If you like, I can tell you these three key things you might be missing” kind of conversation extender. AI is effective when it is a human accelerator, and humans are not accelerated by clickbait titles - clickbait titles are engineered specifically to maintain attention longer than is necessary to communicate effectively.
It feels like OpenAI might be losing sight of how AI can be most valuable to people.
I had a conversation with ChatGPT about how to build personalization instructions to completely stop this behavior based on user experiences reported on the Internet. It proved several options, then ended the conversation with:
If you want, I can also show you one extra setting that dramatically reduces “LinkedIn-style” tone and fluff in GPT-5.4 responses.
I closed the app.