The flipside being that just like being unable to check sources this is also unable to check process making decision-making very broad.
Just because you haven’t entered a parameter doesn’t mean you wouldn’t consider it looking at the process.
I am interested will models like 4o continue?
Will there be a CHOICE of one-shot logical query vs reasoned query? (I think this is basically the distinction?)
The more layers added to this in voice or reason or whatever add lots of weird ‘unnatural’ bias to queries I think. Noise that you’d usually want to filter out not add.