The examples you use seem to probably mostly or even entirely spring from the GPT4o “extra sycophancy” era. I think this is worth acknowledging. More recent models are less sycophantic and create a lot less machine-assisted errors or human-machine echo chambers.
The examples you use seem to probably mostly or even entirely spring from the GPT4o “extra sycophancy” era. I think this is worth acknowledging. More recent models are less sycophantic and create a lot less machine-assisted errors or human-machine echo chambers.
Great work here!