The “conversation mode” (with the white circle) on the ChatGPT iOS app has always been a bit of a mess. It jumps ahead too quickly whenever I pause to think, and holding my thumb on the white circle feels like a clumsy workaround—it ties up my hand and interrupts my workflow. Worse, if I try to “interrupt” ChatGPT while it’s speaking, it often breaks the whole conversation, forcing me to exit and re-enter voice mode.
I know that recently, the pausing problem has improved somewhat - but I’ve still always preferred “dictation mode” —it actually let me handle complex tasks. I could multitask on other apps with ChatGPT running in the background, review draft transcriptions for errors, and even scroll back through older comments while recording, which was crucial for drafting lengthy memos or emails.
But now, a couple of seemingly minor changes have turned narration mode into complete garbage. Tapping the “enter” button doesn’t start dictation mode anymore; it throws me into “conversation mode,” which I really only use for playing games with my kids. Holding the button down to use narration mode now combines the worst of both worlds—I lose all the multitasking and editing abilities I used to have, and I can’t do anything else without being interrupted or cut off.
To top it off, I used to be able to review and edit a draft transcription before sending it. Now, the app just uploads the final transcription as soon as I release the button.
I get that OpenAI might think this “streamlines” the app, but it really doesn’t. This morning, I ended up recording a memo on the iPhone’s Voice Memos app, uploading the file to Otter for transcription, and then copying and pasting that transcription into ChatGPT. This is not streamlining; it’s prioritizing users who treat the app like a glorified Tamagotchi or a virtual friend, over those of us who need it for serious, detailed work.