I am looking forward to a spirited conversation on this opinion piece.
Plus you have to love my “Obsidian rock infused with AI” image.
I am looking forward to a spirited conversation on this opinion piece.
Plus you have to love my “Obsidian rock infused with AI” image.
Sorry, but a huge no from me. Don’t want AI integrated into Obsidian in and of itself, plus it would mean spending precious developer time on this instead of crucial core features.
I think many people consider AI to be a crucial core feature. Or at least, a feature that is currently in the process of becoming table stakes
Totally agree. The idea of the core team making standardized plumbing is a great middle ground, and arguably is more in line with their position on AI features than doing nothing.
The way I see it, choosing not to use AI with your notes is going to be like choosing to write all your notes on paper. Sure, there might be some advantages, and many people will like it for their own reasons. That’s totally fine, but most people will not choose this.
While many others would start searching for another note-taking app if Obsidian jumps on the AI hype train. Integrating LLM into Obsidian is NOT a core feature by any stretch of the imagination. Core features presupposes building upon existing ones, not sidetracking into an entire different lane, one riddled with privacy issues at that, all for the benefit of chatting with my notes. No thanks. Give me dynamic views and other ways of leveraging properties, better tag management and such instead.
“an LLM is not a core feature”
I think in a few years it will be. What constitutes a core feature for a note taking app is pretty subjective; it’s based on the definition people hold of what software should be able to do. I think our expectations of what we expect all software to do will be irrevocably changed by AI. You can’t ignore the internet in our modern society, you won’t be able to ignore AI tomorrow.
Also, are you aware there are local LLMs that have no privacy concerns? If Obsidian shipped with a local llm such that nothing left your computer, would you still have issues or concerns?
I think it will not, both are our subjective opinions. Mine is based on LLM’s insatiable energy and computing needs, coupled with subpar results in comparison. As long as AI has no bloody idea what it is saying, it is anything but a desirable unavoidable computing future. So no, even locally and no privacy issues, I’m not interested in needing a mini-supercomputer so that it can chat with me about my notes. What I want is a note-taking app that enables ME to structure, find and chat with my notes myself the way I need and want. Even to the extent that you want to use LLMs, there’s nothing stopping you from using one of them, without shoving them into every software in existence.
I think several users here are in a bubble that would think AI is a crucial core feature (even though there are still plenty who don’t, obvi). As it stands, though, the general public is probably still not at a point where inclusion of LLMs is crucial or even a consideration. Just search “public perception of AI”.
I don’t believe the inclusion of LLM tools into Obsidian would be something that happens in the near term. Considering Obsidian’s manifesto, I would say that investing developer effort on inclusion of LLMs caters to one way of thinking, which is somewhat at odds with being “malleable” in my opinion (on an aside, I think it’s worth considering who actually powers the AI hype train and whether that is contradictory to Obsidian’s principle of being guided by users, not investors). However, making the means available for users and plugin developers to integrations seems well within the principle of maleability. I don’t know where that would rank amongst whatever else is in the app developer’s backlog, though.