Last week, OpenAI announced they are shutting down the private Slack community for people who had access to GPT-3.
Instead the conversations are moving over towards a public online forum instead.
I’m going to miss the community and was disappointed by the news. I had a feeling the axe was coming from a few things I could tell early on, but I still wanted to participate. Despite this inevitable axing, I still had personally spent a lot of time and invested in the community by helping out where possible, participating in discussion, and sharing interesting content.
I just finished recording a podcast episode questioning the nature of this decision and primarily, how does OpenAI view its relationship to the developer community? Is it a developer community they are building or do they still view what they’re building as just an API business product?
My deep fear is the community evolving into some sort of corporate afterthought. In the same way AWS doesn’t really prioritize their community, so they have these very corporate like online forums. Microsoft and Apple have similar online forum setups as well which aren’t … the greatest. Which is a real shame for OpenAI.
In the podcast, I argue that there was a real opportunity here to build a one-of-a-kind AI developer community instilled with the cultural values OpenAI believes in. Instead, I’m not sure what kind of culture may or may not be created on the community online forums. The Slack community was still nascent and had real potential. Now, it’s just … well … gone (or technically not gone yet, the official cut off date is June 30).
I just finished recording this episode and uploading it but the larger question I think underlying this that I still have - will the community be involved at all with the development of GPT-4 or DALL-E? How much of OpenAI will be top down and not developer community driven?
I think the safety and policy risks are greater without an intimate relationship with the developer community. The slack group brought some transparency to OpenAI, their staff, their decision making process, and now that transparency is gone. Since forums are fully open to the internet public and are publicly searchable (possibly for decades on end), I think there will be fewer honest conversations on there. This is particularly true when it comes to the community proactively raising safety and ethical risks, which are very serious in my opinion when it comes to the deployment of large scale language models. I guess only time will tell, but to be honest, I’m still disappointed about this whole thing to the say the least.
I said in the podcast OpenAI hadn't asked anyone before shutting down the Slack group. I'm now learning this was not true, they had asked a few select people, just not the community at large.
I'm getting interesting feedback from various people and wanted to share some interesting counter arguments:
An online forum brings even more transparency than a slack group (which is semi public anyways)
Many conversations are happening in different timezones, forums will handle this situation better
I'm hearing slack search is not as good or people are not using it enough and sending messages over with very little thought/consideration
You can still DM in discourse which is interesting and something I didn't know
More focused conversations and better search capabilities
I guess one mistake I've made is that I've been too hard in judging the forums too early, but I guess it still depends on how active the forums will be and also on each individual's communication tool preferences.
Also, in fairness to OpenAI, I don't know what they have planned for the forums in the future, it could be really big things.
I guess the main thing is that Slack just feels more intimate to me but maybe I'm wrong and others don't actually feel the same way.