I pretty much agree. I just wrote a comment about why I believe LLMs are the future of rational discourse. (Publishing this comment as comfort-zone-expansion on getting downvoted, so feel free to downvote.)
To be more specific about why I agree: I think the quality/agreement axes on voting help a little bit, but when it comes to any sort of acrimonious topic where many users “have a dog in the fight”, it’s not enough. In the justice system we have this concept that a judge is supposed to recuse themselves in certain cases when they have a conflict of interest. A judge, someone who’s trained much of their career for neutral rules-based judgement, still just isn’t trusted to be neutral in certain cases. Now consider Less Wrong, a place where untrained users make decisions in just a few minutes (as opposed to a trial over multiple hours or days), without any sort of rules to go by, oftentimes in their afterhours when they’re cognitively fatigued and less capable of System 2 overrides, and are susceptible to Asch Conformity type effects due to early exposure to the judgements of other users. There’s a lot of content here about how to overcome your biases, which is great, but there simply isn’t enough anti-bias firepower in the LW archives to consistently conquer that grandaddy of biases, myside bias. We aren’t trained to consistently implement specific anti-myside bias techniques that are backed by RCTs and certified by Cochrane, not even close, and it’s dangerous to overestimate our bias-fighting abilities.
I pretty much agree. I just wrote a comment about why I believe LLMs are the future of rational discourse. (Publishing this comment as comfort-zone-expansion on getting downvoted, so feel free to downvote.)
To be more specific about why I agree: I think the quality/agreement axes on voting help a little bit, but when it comes to any sort of acrimonious topic where many users “have a dog in the fight”, it’s not enough. In the justice system we have this concept that a judge is supposed to recuse themselves in certain cases when they have a conflict of interest. A judge, someone who’s trained much of their career for neutral rules-based judgement, still just isn’t trusted to be neutral in certain cases. Now consider Less Wrong, a place where untrained users make decisions in just a few minutes (as opposed to a trial over multiple hours or days), without any sort of rules to go by, oftentimes in their afterhours when they’re cognitively fatigued and less capable of System 2 overrides, and are susceptible to Asch Conformity type effects due to early exposure to the judgements of other users. There’s a lot of content here about how to overcome your biases, which is great, but there simply isn’t enough anti-bias firepower in the LW archives to consistently conquer that grandaddy of biases, myside bias. We aren’t trained to consistently implement specific anti-myside bias techniques that are backed by RCTs and certified by Cochrane, not even close, and it’s dangerous to overestimate our bias-fighting abilities.