Rant: GPT-4o is just a polished version of what we already had

I spent last weekend testing GPT-4o against Claude 3.5 Sonnet for writing email drafts for my pest control clients. Same prompt, same instructions. GPT-4o gave me this super polished, almost too formal reply that sounded like a robot pretending to be a CEO. Claude nailed the casual but professional tone I actually use. I ran 12 more tests on different tasks and Claude won 9 of them. Anyone else noticing the emperor has no clothes on these new models?

2 comments

2 Comments

uma3062d ago

Robot pretending to be a CEO" - what specific part of the email gave that away?

kelly_coleman612d ago

The part where it said "I look forward to synergizing our core competencies" basically screamed bot. Real CEOs just say "let's work together" or something. My guess is the AI was trying way too hard to sound important and forgot actual humans don't talk like that.