This is Part 2 of me, trying to help improve ChatGPT. Hopefully you are there reading this OpenAI. This one will be short.
- Still, the reader is not where it should be. Half the time when I click read out loud, it just won’t work, and 99% of the time it won’t work with Deep Research. Let me explain how I see this situation and please correct me if I’m wrong, community. OpenAI has arguably a team of the most brilliant people on planet Earth. You guys have created a true masterpiece that it would be hard to live without, now that I use it for so many things. You know how some things in life are so absurd and baffling that it keeps you up at night thinking about it? This is one of those things. If I had 3 wishes from an all-powerful magic genie, I would strongly consider making one of those wishes, "Genie please tell me why OpenAI cannot figure out how to make text-to-speech work.
2)Deep Research is fantastic, consider making a repository that separates/sorts for all Deep Research questions, and filters out everything else. I know you have folders but they are kind of annoying, not entirely sure why.
3)Consider a search bar, to search for the previous question/ dialogues you have previously had. All too often I have an insightful conversation and it’s completely lost. Would it really be that difficult to do?
-
Where is OpenAI on these forums? Where is their collaboration? I want to see them make cool topics that are insightful. I want to learn more about AI. I want them to ask US questions on how they can improve ChatGPT. Not sure how the burden shifted on the consumers to try to feed the providers, the “Change this. This feature doesn’t work. You need this feature.” If I owned/managed OpenAI, the very first thing I would do is have a high level of communication with all their clients, ESPECIALLY THE PRO MEMBERS WHO PAY 200/MONTH & THE PEOPLE WHO MADE AN EFFORT TO JOIN THIS FORUM! Come on guys. There is way too much competition in this AI Arms Race to not be doing the fundamentals.
-
The support team is very kind, 10/10 customer service, but there wasn’t a single time they have actually resolved a issue/concern for me. I completely understand there are things that they cannot control, but they should at least know when to escalate an issue. For example, I took time out of my day to make a topic post about how i lost like 80 deep research questions, I figured that post alone on a forum would be enough to get the situation resolved, nope. Okay fine, OpenAI has a lot on their plate right now, maybe they aren’t reading forums recently. I send the forum directly to the support team (the link) and I get back a response basically reading me the FAQs. I try again, mentioning, I’m familiar with the FAQs and surely there are extenuating circumstances when someone loses 80 deep research questions. Long story short, unresolved, and I am a pro-tier subscriber. This is business 101. How are you going to jeopardize your relationships with your highest paying clients, business 101.
If there is one thing you take away from this. Please make sure it is #6.
- GROW YOUR SALES TEAM. I emailed the sales team asking them “Hey, I want to give you more money is that okay?” and I got some random automated response back saying I would be a good candidate for team enterprise. This is just flat-out unacceptable. What I really said was something like, “I love ChatGPT and I want to share this with more people, allow for me to pay for their membership for a month.” I will break this down since it seems we need a sales lesson. I buy 50 people the Upgraded paid ChatGPT subscription (out of the kindness of my heart), they like it, and they tell all their friends, family, etc, “Hey you need to purchase this subscription, it is a game changer”. This might even be my second question for the all-powerful magic genie.
I wouldn’t waste my time writing all of this if I didn’t care about the future of Open AI, I work 112 hours a week.
NUMBER 7 ALSO VERY HIGH-YIELD
- The current feedback system—asking users ‘Which response do you prefer?’—is limited because it doesn’t allow for nuanced feedback. Responses often differ in their strengths across multiple important dimensions, such as clarity of language, logical layout, emotional connection, overall effectiveness, accuracy of information, relevance to the prompt, level of detail or conciseness, creativity, tone appropriateness, practical usefulness, and engagement. Instead, it would be beneficial to prompt users with, ‘Which response do you prefer in the context of layout, language, connection, effectiveness, accuracy, relevance, detail, creativity, tone, usefulness, or engagement?’ By incorporating these additional dimensions, users could provide targeted, actionable feedback, helping ChatGPT continuously refine and optimize its responses.
I said this was not as short as I wanted it to be, but please, if anyone has contact with OpenAI, help me help them and send this post to them. If I dont get any response on this forum from an OpenAI developer saying something even as small as “received”, im going to find a way to get this to them.
Edit: Just checked my messages from support and they have gotten back to me and they escalated the issue, which I greatly appreciate. I will say, they asked me for a date & time that this occurred which is challenging since I use chatgpt for around 6 hours a day, and I will likely be scrolling in my history for hours. My point on #3 would certainly help with this.
Thank you for your time. Apologies for spelling & grammer.