Custom Conversations: OpenAI's ChatGPT Unveils Personalized Response Guidance

0:00The wait is over. Dive into Audible's most anticipated collection, The Best of 2025. featuring top audiobooks, podcasts, and originals across all genres. Our editors have carefully curated this year's must-listens from brilliant hidden gems to the buzziest new releases. Every title in this collection has earned its spot. This is your go-to for the absolute best in 2025 audio entertainment. Whether you love thrillers, romance, or nonfiction, your next favorite listen awaits. Discover why there's more to imagine when you listen at audible.com slash best of the year. Chat GPT is currently working on an amazing new feature that is going to make a lot of things much, much simpler.

0:46So without further ado, let's dive into it. So this news was actually broken on Twitter of all places by someone named Gabriel Cohen, and he dropped the scoop that OpenAI is working on a new Chat GPT feature called Chat Preferences. So essentially how this works is when they're rolling out new features, they select small groups of people to try it. So I think it's right now in the alpha phase. It will then go to beta phase and then it will go, you know, out to everyone. So he recently just got a notification when he opened up ChatGPT that said introducing chat preferences. Chat preferences let you share anything you'd like ChatGPT to consider in its response.

1:25Your instructions will be added to new conversations going forward and you can edit it and delete them at any time. This information will also be used to improve model performance and shared with plug-in developers And then you know you can visit their little help center to learn more and it allows you to click ok to accept that now If you accept that Essentially users will be given the ability to add context about themselves as well as their preferences for how chat gpt should respond So this information is going to influence all chat conversations now this is something that they actually recently announced a feature they were adding to their API and not everyone knew exactly what it was going to be for but I think what's interesting is the fact that this was included in their API and that is indicative of the fact that they are building this straight into their actual chatgpd model that everyone is going to be using not just the API that developers use and I think that makes a lot of sense so essentially you can toggle on chat preferences once you've toggled it on you have the ability to fill out one of two boxes so The first one asks you, what would you like ChatGPT to know about you to provide better responses?

2:29So you could fill this out by saying, I am a lawyer, I'm a doctor, I specialize in X, Y, and Z thing, I'm in the marketing field, whatever, you know, you think would be the best, whatever your use case is. The next thing it says is, how would you like ChatGPT to respond? And you can actually fill that out. What's interesting here is, you know, if you use this for a specific job or specific task really repetitively You can actually put all of those instructions in here and you know, it's gonna do that every time So for example, if you're a journalist and you're constantly using ChatGPT to help you You know find sources or perhaps read a massive document and pull out some useful quotes You could have that, you know, put in here ahead of time and you just paste in the Paste in the giant document and it would just pull out the quotes for you every time because you've already written the prompt and you know, preemptively put that in there.

3:16So some examples of how they intend for it to be used. They give a couple examples of what they call our thought starters. So you could say where you based, you know, what you do for work, you could tell it what your hobbies and interests are, you can tell what subjects to talk about. And you can also say what some of the goals you have are. So by kind of giving it this, this information, you could say, I am an aspiring journalist that hopes to someday win a Pulitzer Prize. I like investigating companies in X, Y, and Z sector. I'm really passionate about X, Y, and Z thing. I'm based in this city.

3:48My hobbies and interests are all of these things. And, you know, this is what my main goal in my career is. You put all that information as a preface. And every time you ask it a question, which could be a regular question, right? You could still say like, hey, help me write X, Y, and Z thing for this Pulitzer winning article. And it could help you to compile something, but it knows all of this stuff ahead of time about you and it kind of has a broader picture of where you're going as like a person where your goals are I think this is really really interesting it also asks how do you want to be addressed so you can say you know how formal or casual chat GPT should be personally I typically am looking for a specific style of response from chat GPT and sometimes it honestly is a little too formal for me so I think this would be really interesting to be able to say you know hey respond in a casual tone about x y and z or like explain this topic to me in a simple way and i you know oftentimes will paste in really complex um ai kind of information and i have it explain it to me a simple way so you could also tell it how long or short a response should be so you could say hey you know when you're responding to me make sure that you're giving me really brief responses that are less than a paragraph or that are only two sentences long or you know so like if you're using it and trying to get the same type of result consistently out of it you can add all of those things you can also um tell chat gpt to have opinions on topics or you can tell it to remain neutral right you can say i really don't appreciate it when chat gpt tries to lecture me about x y and z topic or i don't like it when chat gpt has the assumption that i believe x y and z and you can put that in there you could probably i mean i don't know but you could probably put your ideological beliefs in your political beliefs you could probably put your religious beliefs anything that you feel like is like a framework for what you believe or how you think i think would be really cool to be able to put those into chat gpt um into this kind of personalization aspect and now when it gives you responses it takes into a it takes into account you know your beliefs and your ideologies because at the moment you know one of the big complaints from people using chat gpt is the fact that it was trained by scientists, it was trained by people, and all of those people have their own beliefs, their own biases, their own thoughts.

6:00You know, even the data that was scraped off of the internet to help create ChatGPT has its own thoughts, ideas, and beliefs. And, you know, inevitably, people are going to disagree with many of those things. So it's kind of cool that you can kind of get it to put in what you think, what you believe, and get responses along that vein, if that's what you're interested in. So as I said earlier, it's currently in alpha. They're going to move it soon to beta and then to GA. So it seems like this is going to be really interesting. It looks like this is going to apply to chat GPT, including GPT-4. Also plugins and also code interpreter, which I think is going to be really interesting.

6:37And I think that it is likely using the system message, which has been available in the API for quite some time. And so this is going to make the instructions have a lot bigger impact than if you just pasted them at the beginning of each chat. And also, I don't know if these the things you'd be putting into this are things you necessarily would be putting into every single chat. Because I think what's cool is here you can kind of put information about yourself, information about your goals, information about your beliefs. And when it gives you a response, now all of a sudden it's like, and even your tone and like how formal or casual you want it to be.

7:07Like there's so much that you could do to get this thing to respond in a way that, you know, meets your personal brand or your company's brand, right? Like this would be really awesome for companies that are like, hey, we always have a super formal tone when we respond. We always, you know, want to make these things look as professional as possible. You can put all of that in to make sure that no matter what it's doing, it's always going to be matching exactly what you want. So I think that is super interesting. I think this is going to be a really exciting feature that actually will have some big implications.

7:37And so I think this is currently in beta. The feature is only available for Plus Plan users and for people in the EU or the UK. I don't think this is yet an option. So users can find the feature in the web settings under beta features or on iOS under new features. I think it's worth noting that any information used in these responses is going to be used, like I mentioned earlier, to improve OpenAI's API model's adaptability. So as the company explains, this is about teaching the model how to adapt its responses to your instructions without overdoing it. So users who prefer not to have their data used in that way can also opt out of this in the data control settings.

8:17So one last thing I will mention about this new feature is that it appears the effect kicks in from the subsequent session with a character limit of around 1 ,500 words for responses or 1 ,500 characters for responses. So that's important to know. But overall, I think this is going to be a really important update. I think this is going to make it a lot easier for many companies and users to really personalize this to what they want it to do. So this is going to be really interesting to follow and see how this feature rolls out when everyone has general access to this and moves out of, I think, more of a beta what it's in right now.

8:50And so this will be something we follow in the future.