How much can Grokbot’s behavior actually be changed through instructions?
Has anyone experimented properly with this
My Grokbot keeps defaulting to a very corrective / moralizing personality and disagrees with me way more than I expected. It’s becoming a real pain in the ass !! Even on normal brainstorming it sometimes feels like it’s trying to steer the conversation instead of just helping me explore the idea.
I’m wondering how much of that is the underlying model vs the bot instructions?? Has anyone found a good instruction setup that makes it:less preachy and more neutral? and willing to explore unconventional ideas?