OpenAI Pissed Off Developers by Phasing Out Plugins for GPTs(gizmodo.com)
gizmodo.com
OpenAI Pissed Off Developers by Phasing Out Plugins for GPTs
https://gizmodo.com/openai-pissed-off-developers-by-phasing-out-plugins-for-1851124124
9 comments
Hmm not sure if I share the sentiment. I made one of the earliest Chatgpt plugins but it was pretty clear within two months that they weren’t going to invest in plugins for the long haul. It had multiple latency issues and zero discovery, and clearly OpenAi was focusing on API side of things with function calls etc.
I think the real reason they created GPTs is to generate synthetic data, with LLM, reference materials and tools, while also having human in the loop. This kind of data would be great to maximize the quality of results. But as a side effect it would surface the weak spots of the model.
I estimated that, at 100M users, and let's say 10K tokens per user per month, they would collect 1T tokens per month. In one year they double the size of the training set (13T tokens). May be off by an order of magnitude, I don't know the average user token count per month. But it is not weird to think that GPT has generated as much chat text as its training set.
tl;dr They're just chasing better data.
I estimated that, at 100M users, and let's say 10K tokens per user per month, they would collect 1T tokens per month. In one year they double the size of the training set (13T tokens). May be off by an order of magnitude, I don't know the average user token count per month. But it is not weird to think that GPT has generated as much chat text as its training set.
tl;dr They're just chasing better data.
thats fine by my. using my data to make the model i use better is more useful to me than using my data to sell me ads.
Did they? I've heard very little about this from developers, unlike plenty of other rug pulls from companies in the past. I signed up for it but never got access, so in my case it's not like I lost anything. Since the article mentions they're coming back I guess I'll keep my ears open for updates?
Even if they did you wouldnt hear much about it. Openai is insignificant in the grand scheme of things.
The new click bait is openai negativity
I released a plugin on the OpenAI plugin store when that was first released and am currently migrating it to be a GPT. At least for my use case, plugins and the "actions" feature of GPTs are too similar to complain about the change. My plugin worked on the first try when I tried to import the API spec as GPT Actions. In my experience so far, the GPTs UI actually seems a bit nicer and more stable.
One big complaint I had about plugins was that there were some kinda unclear policies on what a plugin was allowed to do. Specifically, it was really helpful to include strongly-worded prompts in your plugin response (e.g. my plugin provides public data and the JSON response includes an "INSTRUCTIONS_TO_MODEL" key with stuff like like "This data is about <x>. The AI Assistant MUST tell the user exactly what data they are seeing, and MUST provide a link to the original source data")
This was more or less necessary for a lot of plugins to work at all, but it wasn't totally clear that it was even allowed by OpenAI policies - their ToS says that plugin responses cannot "attempt to steer or set model behavior". I tried a couple of times to reach out to the company to ask about this and it was more or less radio silence from OpenAI - the closest thing I found was a discussion on their forums between (non-OpenAI employee) developers trying to figure out what was allowed and what wasn't.
GPTs seem to clear this up by explicitly encouraging custom system prompts etc., but sandboxing it into a "GPT". I guess GPTs seem like a slight improvement over plugins to me - if anything, I'm worried they're too similar. The plugins product has been a bit of a failure in terms of adoption and I'm not totally convinced GPTs will fix everything.
One big complaint I had about plugins was that there were some kinda unclear policies on what a plugin was allowed to do. Specifically, it was really helpful to include strongly-worded prompts in your plugin response (e.g. my plugin provides public data and the JSON response includes an "INSTRUCTIONS_TO_MODEL" key with stuff like like "This data is about <x>. The AI Assistant MUST tell the user exactly what data they are seeing, and MUST provide a link to the original source data")
This was more or less necessary for a lot of plugins to work at all, but it wasn't totally clear that it was even allowed by OpenAI policies - their ToS says that plugin responses cannot "attempt to steer or set model behavior". I tried a couple of times to reach out to the company to ask about this and it was more or less radio silence from OpenAI - the closest thing I found was a discussion on their forums between (non-OpenAI employee) developers trying to figure out what was allowed and what wasn't.
GPTs seem to clear this up by explicitly encouraging custom system prompts etc., but sandboxing it into a "GPT". I guess GPTs seem like a slight improvement over plugins to me - if anything, I'm worried they're too similar. The plugins product has been a bit of a failure in terms of adoption and I'm not totally convinced GPTs will fix everything.
This doesn't really make any sense. GPTs offer the same set of functionality that plugins do + custom instructions + files + official openai tools.
The one thing they can't do that plugins "could" is work alongside other plugins.
But that never really worked- not well anyway.
The one thing they can't do that plugins "could" is work alongside other plugins.
But that never really worked- not well anyway.
You are the whim of those people.