Steve Strickland: GPT-5 is the first model I’ve used that will deliberately give a wrong answer to ‘check you’re paying attention’.
This fundamentally unreliable technology is not going to put us all out of work.
Is that intentional, or is the name a hallucination and the cheeky comment an intervention aimed on exploiting the fact that models can tell when they are hallucinating.
There was already Cursor AI who decided to refuse to generate code and to offer a paternalistic justification. And now GPT-5 inserted a joke into an answer and claimed to check that the user is paying attention. Is there already a way to reproduce this effect? Does it mean that GPT-5 and Cursor AI tried to be aligned to the human’s long-term interests instead of short-term sycophancy?
Is that intentional, or is the name a hallucination and the cheeky comment an intervention aimed on exploiting the fact that models can tell when they are hallucinating.
There was already Cursor AI who decided to refuse to generate code and to offer a paternalistic justification. And now GPT-5 inserted a joke into an answer and claimed to check that the user is paying attention. Is there already a way to reproduce this effect? Does it mean that GPT-5 and Cursor AI tried to be aligned to the human’s long-term interests instead of short-term sycophancy?
EDIT: Alas, I tried to do this experiment (alas, with a model more primitive than GPT-5) and received this result, which is a joke instead of the world map.