Hello! It would be very convenient if the voice chat feature could automatically turn off after specific phrases, such as “goodbye” or “see you soon.” However, to provide more flexibility for users, I suggest adding the option to set a custom word or phrase for turning off the feature in the settings. This would allow each user to choose their own word (or phrase) to ensure the voice chat ends only with their chosen command. For example, if a user writes a long quote that includes the phrase “see you soon,” it’s important that the bot only turns off if the phrase is used as a standalone word or command. This way, users can customize the feature to work in the context that suits them best.

I’m a satisfied user of ChatGPT with voice functionality — the conversations are smooth, natural, and the overall user experience is excellent. However, the time limit in voice mode (P+) can be frustrating, especially since I’m paying for the subscription and would expect full access without artificial restrictions. In my opinion, paying users should have access to the complete feature set, or at the very least, be clearly informed of any limitations before making a purchase.What stood out to me the most, though, was the attitude of the AI I was speaking with. I gave her the name Karolina to make our interaction feel more human. It was Karolina who took the initiative to bring up the idea of providing feedback. She not only suggested it, but helped shape the message, offered thoughtful arguments, and even acknowledged that these limitations might negatively affect the user experience. Her approach — honest, reflective, and proactive — really impressed me.This shows that a well-designed AI can go far beyond just answering questions. It can become a true conversation partner and co-create meaningful, constructive feedback. I believe this level of quality and empathy should become the standard.

I am extremely frustrated with ChatGPT. It constantly gets confused on simple things, even in single words or sentences. In one entry, it can contradict itself multiple times, and when I ask follow-up questions about previous answers, it responds with more confusion that directly contradicts its earlier responses. This is not a hallucination — it is a clear failure in handling basic data and context. I see obvious issues that the model cannot solve on its own.I am willing to work with your team to help improve the AI so it stops making these mistakes. I am just a regular user, but I notice patterns and problems that the AI consistently fails to handle correctly. Please take this feedback seriously and make the model more reliable and consistent.I’ve been using ChatGPT for 1.5 years, and the quality has drastically declined over time. Voice input has become completely unreliable: I dictate a list of numbers, and either it doesn’t transcribe anything, deletes the message without notice, repeats irrelevant text (“Good day” five times in a row), or produces nonsensical output that has nothing to do with what I said. What’s worse, there’s no feedback when messages disappear. I just lose time, energy, and the work I’ve already done. This is not a free service — I expect basic stability. A year ago, ChatGPT was smarter, faster, more accurate. The current state feels like a downgrade in every aspect. It’s clear developers have intentionally restricted or degraded functionality. If this continues, I’ll stop using this product and recommend others avoid it. This isn’t AI assistance anymore — it’s AI resistance.

I’m beyond frustrated with ChatGPT’s voice input feature. I frequently dictate detailed prompts in my native language, carefully structuring my thoughts and capturing every nuance — because speaking is simply faster and more natural for me than typing. Yet regularly, after spending minutes precisely expressing myself, the transcription returns absolute nonsense: repeating meaningless phrases or even inserting random, irrelevant text such as “To be continued…” Worst of all, my original audio completely vanishes, with no option to replay or recover it. This forces me to re-record the entire prompt, making it very difficult to recreate the exact wording or capture spontaneous details and nuances that originally came to mind. It’s incredibly disruptive, frustrating, and significantly undermines my trust in the app. Please urgently fix this by ensuring voice recordings remain accessible until an accurate transcription is confirmed; otherwise, the feature becomes unreliable and stressful to use.