As a heavy user, I feel that when GPT-5 was first launched, its anthropomorphic system might have been down-weighted due to cost or technical reasons. Within just two hours of its release, I realized its level of logic had improved a lot, but its tone felt similar to the 4o model where I had explicitly turned off the anthropomorphic system. Recently, however, it seems to have improved significantly.
> If so, why did OpenAI give it this personality?
Asking "if so" about a speculated or perceived behavior is confirmation bias.