Tried both Claude and GPT-4 for drafting my newsletter and Claude was way better at getting my tone right without edits

Been running a small community newsletter for about 6 months now, just local stuff. I spent 3 weeks trying to get GPT-4 to sound like me but it kept coming out too formal or weirdly salesy. Switched to Claude last Tuesday and my first draft needed maybe 2 small changes. The difference was pretty obvious once I compared them side by side on the same topic about our farmers market schedule. Has anyone else noticed one model just clicks with your writing style more?

2 comments

2 Comments

danieljenkins8d ago

So I guess the robots just have personal favorites now too.

stella_shah898d ago

Personal favorites" is a strong way to put it, daniel. But it does make me wonder - what exactly happened that made you say that? Like, was it something small that just felt weird, or did the robot flat-out refuse to listen to someone and then do the same thing for the next person? Because I've seen this kind of thing come down to bad training data more often than actual favoritism, but your mileage may vary if it's a newer system. I'm genuinely curious what the specific situation was.