Stephan Go was interviewed by Casey Newton for the Verge's Decoder Podcast.

On productivity tools needing community more than AI -- or rather, how Steph thinks about adding AI to Obsidian -- he posted this screenshot from the transcript with highlights, roughly as an answer to What is the Obsidian view on AI and productivity tools? Will you add features like that?:

Our philosophy as far as how it would ever make sense for Obsidian is that it has to fit with the principles that are in our manifesto, which is that it would have to be private. We’re not comfortable with the idea that our users’ data could be stored in OpenAI servers without their consent. I think a lot of tools out there are just kind of defaulting to this feeling that there’s an arms race. We’ve got to put AI into everything. Let’s put a little magic button everywhere. I don’t think that’s us. We want to give users confidence that their thoughts are theirs, that things are not going to be used to train the next LLM.
That said, I do think AI can be really powerful for certain uses. So the question is, in the long term, do we end up giving an API to the plugin community so that they can build those types of functionalities more easily? Right now, we’re not working on it. We’ve been holding off and watching what’s going on. We don’t feel a sense of urgency to suddenly put all these things in there because, to be honest, the plugin ecosystem is there for you and you can do it if you really want that. There are things much more important to us on the priority list that we want to work on first, that we would rather set our time aside for with our limited capacity.

Elsewhere he talks about the Obsidian plugin system and how many emerging popular plugins are all about AI.