Contact ChatGPT Support

by Felixlunzenfichter

No doubt, this technology is absolutely life-changing, especially for me. You don’t know how thankful I am. However, I’m not reviewing the underlying technology, I’m reviewing its implementation on iOS. I use it for about 12 hours a day for my work, so I’m a bit picky. But it seems to me like you should have the resources to do a great job here. 1. Why does the app not work in the background? If I send a request, I have to stay in the app to wait for it. That’s just horrible, and it’s completely unnecessary because you’re doing everything in the cloud anyway. Just finish processing and send a notification when you’re done, especially for browsing features. I don’t care if it takes 10 minutes to find the answer. I’d be glad if it’s browsing in the background. (To browse in the application as of May 24, start a conversation in the browser, select the browsing option, then switch back to the app. Browsing is still in beta, and you cannot select it directly in the application, which is fine.) 2. Please add the option to choose Chat GPT4 as a default. I never use GPT 3.5, and it’s annoying to have to switch every time. 3. Don’t let the speech-to-text button disappear once there is any text. If there is any text and speech recognition is started, just append. (The speech recognition is completely amazing, congratulations!) As soon as those features are ready, I’ll be glad to give five stars. Thank you very much.

by Cjjcjfdhshhshddhdnd

The most recent update has basically made the bot extremely soft and sensitive in GPT 3.5 mode. When you try to write a story even with very innocent topics and comical topics it refuses claiming it’s very inappropriate and that it can’t do it at all. It also refuses to answer questions mainly about historical subjects it thinks is remotely controversial such as battles and military history. It also gives inaccurate information even when it does work or it repeats information especially when you ask it to list off something. It also sometimes ignores your instructions when trying to make creative stories by adding its own version and not following your instructions along with sometimes when you ask it to give info and omit obvious info or info you don’t need it’ll give you that anyways such as asking it to omit a topic or event from a list but it still gives it anyways. Although gpt 4 seems to work well and is quite good, the free gpt 3.5 is extremely useless and can’t do much right not even simple math problems or data widely available on the internet for years now. Please fix this as this tech has proven to be very useful in previous versions such as gpt 3 and the newer gpt 4 also seems to work but this gpt 3.5 is extremely frustrating useless and frankly ridiculous to use. And I’m not forking over 20 bucks a month just to have the bot somewhat usable. Thank you.

by Daniel778755

Initially the new voice mode did very well, but since the first release a mere few days ago, the chat quality has deteriorated dramatically, such that it is not worth the plug subscription anymore. We all knew OpenAI reduced the quality of the text responses, including a noticeably worse performance in coding, reasoning and literature, but OpenAI has also applied this same policy now to their voice mode models.

I think their marketing strategy is clear by now, for those that know how marketing works. Initially they release the most capable model to entice users, e.g. people posting amazing reactions online and engaging many viewers. Then very shortly after they start reducing the capabilities of the model to save costs. Because OpenAI’s product in ChatGPT is so versatile and difficult to comprehend and track over time, they can just reduce the quality of the model and save tonnes of costs without most people noticing. It is an intractable product, with many different facets and versatility in responses, that makes it nearly impossible to track.

However, I have taken screen shots of my first interactions with voice mode, where the AI confirmed it could guess the tone in my voice, and guessed it correctly on multiple occasions. Now when I ask it, I get a negative response that the model cannot guess my tone.