31–40 of 132 posts
I suppose that up/downvoting a bot to give it more "score" somewhat would be based on people voting individual answers inside that GPT.
As others in the tech scene have said, it might be hard to find the real valuable GPTs since the barrier to build is so low. At least with the app scene, learning Swift/Javascript/ObjC created a barrier, with ChatGPT store anyone and everyone can create a GPT.
It's actually not that low. GPTs allow external functions and the moment you want to do anything interesting you actually need to be able to build it. For example, if your GPT is going to design and host a webpage for you then you will need to be able to create a service that can talk with that gpt, take its input and deploy it. Other GPTs that don't require a hard skill might also turn not to be trivial due to promp…
> It’s been two months since we announced GPTs, and users have already created over 3 million custom versions of ChatGPT.
That's a really low barrier, especially considering that only paying subscribers have had access
Interestingly, I asked the exact same question to the regular ChatGPT and it correctly recommended the nearby walk.
Earlier quoted context omitted.
It's actually not that low. GPTs allow external functions and the moment you want to do anything interesting you actually need to be able to build it. For example, if your GPT is going to design and host a webpage for you then you will need to be able to create a service that can talk with that gpt, take its input and deploy it. Other GPTs that don't require a hard skill might also turn not to be trivial due to promp…
From the announcement: > It’s been two months since we announced GPTs, and users have already created over 3 million custom versions of ChatGPT. That's a really low barrier, especially considering that only paying subscribers have had access
A couple of examples would be enough.
It reminds me a bit of the time when widgets were popular on Facebook or other sites, and I have mostly a negative memory about it.
I mean, apparently most of it is free, but I guess they store the data and who knows if they might then sell it to data brokers.
Earlier quoted context omitted.
From the announcement: > It’s been two months since we announced GPTs, and users have already created over 3 million custom versions of ChatGPT. That's a really low barrier, especially considering that only paying subscribers have had access
IMHO "barrier to entry" implies success and not mere existence. It's like barrier to entry into singing, everyone can sing but the barrier to entry into singing career is actually quite high. Very few things have true barrier to entry and its mostly about purchasing an equipment or license.
The music industry existed long before the internet came on the scene so there are many avenues for distribution and discovery. GPTs have been sharable for a few months now and the market looks abysmal with every website/catalog/database of custom GPTs being flooded with mediocre prompt engineering.
Hopefully the official GPT store overcomes that but I'm not very confident - one of the top GPTs listed is Grimoire and a few others and I don't have a high opinion of it after trying to use it over the last few weeks.
If you haven't see it already, there's Rabbit r1[0] which takes the "in the future there wouldn't be individual apps" approach and on the other hand OpenAI seems to adopt the "in the future, we will host all the apps" approach. I'm curious who will win this one but IMHO there's a value in having an individual well defined app even though I previously argued that in the future there wouldn't be individual apps. The va…
What is going on with their rollouts or A/B testing or whatever they're doing? They have this big announcement, and yet my paid account doesn't show this functionality at all.
https://chat.openai.com/gpts doesnt load for you?
OpenAI has implemented controls to stop the model from adding hallucinated parameters to an action payload... but this results in user-facing failures.
I initially worked around the user-facing failures by wrapping the entire payload in a {"request": {... payload}} structure (which helps because the controls only perform a shallow check ).
It is frustrating that users have no way to view the action response, even though users can view the action request. Not infrequently, the model will take an essentially empty or irrelevant response and silently ignore it, hallucinating an answer as though the response had been informative... so it's hard for users to trust what they see in the generated output.
It would be so easy to enable a toggle for users to inspect the response, but I think the OpenAI team wants to somehow "protect" the IP or internal decisions of custom GPT "creators. It would at least be nice to have a toggle for developers who don't feel proprietary about those details. And maybe a fork button :-)
Earlier quoted context omitted.
That (fallacious) argument applies to nearly every other "tech" platform, from YouTube to Hacker News. The barrier being low is not the problem -- the question is how high does the ceiling go. Is this thing capable only of preconfigured prompts based on a customized data upload, or something vastly more?
Yeah and Youtube is absolutely swimming in garbage. Without dang, Hacker News would be too.