This seems like a genuine innovation in outlining as idea processing, and perhaps a taste of the future. Roam Research now has about 60 available plugins, and one of them, called "Roam AI", implements the OpenAI engine (you have to supply your own API key). At the end of any block, you can type "qq" and then choose from a few different response types (including "devil's advocate"), which will then be automatically generated in the subsequent block. Initial testing suggests this could be a powerful tool. I've found very little documentation, and I don't have a link to share (you can instlal the plugin from within Roam), but here's an example:

I wrote:

I think outlining is a powerful way to organize and process ideas in order to generate clarity and insight.

And here is the Roam AI devil's advocate response (the fulll three-part response was automatically generated in one go):

- 1) Outlining can be too rigid and formulaic, stifling creativity and originality.
- 2) Outlining can be time-consuming and tedious, taking away from the actual writing process.
- 3) Outlining can give the false impression that the writing process is linear, when in fact it is often messy and non-linear.

Ha! That's some clever AI.

I guess I've been out of the loop. Turns out there's already been a Logseq plugin that does this for a while:

https://github.com/briansunter/logseq-plugin-gpt3-openai

There's also a similar Obsidian plugin built on Open AI

https://github.com/nhaouari/obsidian-textgenerator-plugin

I don't see the point of writing if a bot does it for you. Sort of like having a bot that chews your cud for you.

Interesting. It hadn't occurred to me to use the plugin to write for me. I thought the devil's advocate mode, in particular, could be helpful in providing counterarguments to one's current line of thought when using outlining to process ideas. Although of course it's still early days.

(Meant to add: The counter-argument approach reminds me of Neil Larson's Maxthink school of outlining.)

Lucas wrote:

I thought the devil's advocate mode, in particular, could be helpful in
providing counterarguments to one's current line of thought when using
outlining to process ideas.

Yes, it does sound useful for brainstorming purposes or to be able to consider perspectives or arguments one hasn't thought of.

Are there any privacy implications to using this with Roam?