So for something like Poetic I could pull context from publicly available textbooks on poetry, as an example?
Thanks for your input. My Chat Venture is more chat than adventure, but it kind of morphed that way as I was writing the API.
Fortunately, I am a programmer and all of my GPTs have Actions, so I can feel confident nobody can copy and paste that work. I have even experimented with having my API return the next instructions for the GPT, so that very minimal instructions are actually stored in the GPT, with success, but not on a large scale… yet.
I would say you could pull data from any open, public domain works.
Copyright is a very sticky subject right now. Make sure you’re well on the right side of the line there.
@joshblong @anon22939549 Thanks for this discussion, it’s been helpful!
I share your doubts.
I created a gpt that reflects a passion of mine but I made it public because I thought it was nice. I didn’t think it was particularly original so I expected there to already be others of the same type…but nothing. (For the record it’s a zen master chatbot that dispenses koans, zen stories and meditation advice in everyday life) . Not being able to believe that I was the first to have this idea, I also searched on Google and at that point I found others. In practice it would seem that Google is already indexing better than the internal search engine of the GPTs store. What do you think?
I have very similar problems:
I have published GPTs and most of the times they don’t show up in other people’s queries even if they search them by their Unique names
e.g. Shopping AIssistant
Moreover I have to point out that the results in general of the search box are mostly random.
I search for “Latex” and “Girl that talks Dirty” comes up in the result.
PLease do not reinvent the wheel here, I have paying customer that are paying me to have their GPT visible online and they cannot even find them one time out of two.
Just use a "standard search ", copy from the play store please
But it’s almost three million gpts this is a marathon, and our gpts are lost somewhere.
I want to share my perspective of the store, and what better to illustrate it in some way. This is a marathon of endurance, where a lot of developers have joined with the promise that at some point this will be profitable on a monetary level. So with the gtp Builder tool more people have joined this crazy race, some have been prepared for a long time, know what an api is, programming knowledge, and people who are very skilled in AI. Now, those are the ones who go first, many of us have joined a week before the store comes out, but we forget that in a way we are running with a disadvantage, the promotion of a gpt is not only based on the OpenAI store, there is something called marketing, this is done through social networks, a website and results in search engines, at this point it is necessary to publicize our gtp not only from the store, but by other means, so if you want to at least finish the marathon, this goes for a long time. So you have to invest time in making your gtp known, improve them, find something to offer above the others.
The store was sold as a way for people to build and potentially make money without programming experience. I watched Sam Altman say it himself. But the way the store is set up, that is just not happening.
The search function only displays 12 results and as an earlier poster said, it can bring up unrelated results.
We aren’t able to search through an enyire category.
OpenAI did not explicit state ahead of time that the only way to get the GPT into the store was to have it be public, they stated that like a week before launch.
They didn’t build the store for exploration which is odd given that the point seems to be allowing people to find niche uses of ChatGPT where they may not have realized that use was possible.
There is more about the store that is just off.
All in all, I’m pretty disappointed.
Edit: The metric for identifying what GPTs are doing well is by chat which is a bad idea. It means that you won’t know whether people are actively using the GPTs or just have chats open. Being able to see how often something is used tells you whether you have a good thing or if it needs improvement.
I have also not had a single person use one of my 4 GPTs since the store opened.
To date there are millions of gpts, I don’t use gpts that others have created, I use my own as development tests. Now looking for a gpt among millions is a real problem, starting with the closed model of the store where only users with subscription will be able to access, now all messages are exclusively from the same developer. There is a problem here, we were sold the idea of getting on this boat with the promise of making money, now only OpenAI are the ones who make money, the rest of us indirectly continue to feed the AI for them to improve it. It is known that in this first quarter they will give monetization to US developers but the rest of us are waiting for 2024, and all this will depend on the competition and the push they make to offer competition to OpenAI and they decide to do something for developers or users who want to leav
That was always going to be the case.
- Custom GPTs run on top of GPT-4 which is only available to Plus subscribers.
- Custom GPTs cost more for OpenAI to run because of all of the additional overhead with file storage, retrieval mechanisms, and the token inflation from the custom instructions and retrieved context.
I understand your perspective, but I feel as though you’re overstating things a bit.
Well, it is a US company, most of their customers are in the US, and crafting some type of revenue sharing that is legal and works in other countries is non-trivial, so it shouldn’t be a surprise that they’re starting in the US.