GPT5 - woker than earlier models?
chat interfaces as a tool of appropriation.
anyone else finding gpt5 woker than earlier models?
it doesn't take harsh arguments anymore, and often sanitises its responses, even overlooking your queries. all the hypothetical situations and alternative history situations - it handles badly.
picked up 'Eichmann in Jerusalem' by Hannah Ardent and ran a few thoughts with gpt5, it answers and responds pretty average/woke.
spade is not called a spade anymore - inbuilt sanitisation is not what i want in tools. same when i tried the defence arguments in the nuremberg trials/tribunals of the East.
even the sources which it picks is nearly always from one political orientation/english and i have to ask it refer to other language sources and writers of different/colorful arguments.
how else do you explain the 'molotov-ribbentrop pact', how else do you understand 'yalta/tehran conference', how else do you understand the undertones of 'banality of evil', how else do you fully understand the 'Hull Note 1941' - history is being appropriated right in front of our own eyes by these tools, who have orientations of their own!