By skyforbes It is no news that since its launch 3 Pro has been criticized for poorly following instructions, or making arbitrary decisions.
That's been a huge disappointment since it's my main use, but it also gives me headaches when it works relatively well.
That's been a huge disappointment since it's my main use, but it also gives me headaches when it works relatively well.
I'm writing a psychology book, and sometimes I discuss ideas with the AI. With 2.5 I had good contributions, it felt like, even rambling a little, I maintained a "common sense."
3 Pro is just chaotic. Jumps to conclusions and stubbornly follows them, does not collaborate or seems to understand the focus of ideas, like 2.5 does. Furthermore, its tendency to omit instructions extends to omitting information, and many times it only analyze half of the text and leave everything else undeveloped.
In short, it has gone from being a good collaborator to an erratic and inefficient tool. I've changed my methods several times, given it many chances, but it seems to be a root problem. I am deeply disappointed in this model.