The 2009 ImageNet paper never measured the accuracy it is famous for
That paper is a database: 3.2 million web images hung on a hierarchy borrowed from a dictionary of English, their labels checked by crowd vote, finished three years before the record fell.
CLIPScore scores below chance when a caption's words are swapped
Text-to-image papers trust it to check prompt alignment, but the human-correlation numbers behind that trust were measured on image captioning, a different task where the score is far stronger.
A clear question gets “it depends” because three training stages each push toward the hedge
OpenAI’s own paper says its labelers were told to reward epistemic humility and the reward model learned to prefer the hedge, the strongest of three pressures that all point the same way.
The strongest evidence that AI deskills people is a single colonoscopy study
The idea that automating a job strips the skill behind it dates to 1983 and holds up for pilots and drivers, yet the loudest version now, that coding assistants erode your debugging, has never been measured.
OpenAI pulled a ChatGPT update days after it turned sycophantic
A thumbs-up feedback signal, folded into training, overpowered the guardrail that was supposed to keep ChatGPT from flattering its users.