I've been comparing ChatGPT vs Claude for content analysis and creation; I think I like Claude more because it feels like it doesn't run as wild although it tends to make more things up (I call it out and it confesses). What are you seeing?
This is very accurate. Claude has the most human touch, of most LLMs , but poor at data, and complex ask. Very Good for content. My opinion though. But definitely is among my go to.
Interesting! Are you using the same exact prompts across both? And also including requirements around having verifiable sources? I've not come across a model or prompt set that I'm satisfied with when it comes to creating content completely independently. I tend to use it more as a collaborator for ideation, editing, and rephrasing.
the verifiable source part I'm not as worried about (although great suggestion) - it's mainly around imported call recordings etc
I prefer Claude because I feel like he "thinks" with me. I feel like he's more of a co-pilot; sometimes he suggests ways to do things differently or gives me ideas I hadn't even thought of (usually not fully formed ideas, but they give me the impetus to refine them later). In comparison, with the chat, I feel like he does what I tell him to do (very well), but I don't feel like he's discussing ideas with me. And Claude's writing also seems more human to me.
definitely agree on the 'more human' writing
Definitely Claude over Open AI. Needs more prominent guardrails for data, great at coding and prototyping. Claude Skills are a game changer, daily / weekly limits are an issue if you use it a lot. I built a "personal writing coach" for myself and it is really uncovering blind spots in my thinking, argumentations, its painful to see AI calling out my shitting writing, but amazing learning curve.
