The serious case for AI welfare never claims the machines feel
A 2024 report by Long, Sebo, Chalmers and colleagues claims only that the possibility is real enough to prepare for, and the leading scientific test for machine consciousness finds no current system that meets it.
A model's "37% hallucination rate" was its wrong-answer share on SimpleQA
SimpleQA keeps only questions a strong model already missed and grades a declined answer apart from a wrong one, so the share it marks incorrect was never a measure of how often a model makes things up.
Ask a chatbot for a random number and it says 7
A chatbot has no random-number generator to reach for, so it draws the answer from a probability distribution learned from human text, which is lopsided the same way people are.
The diffusion paper behind today's image generators could not take a prompt
Ho, Jain, and Abbeel's 2020 paper reached a state-of-the-art score on the 32 by 32 CIFAR-10 set by learning to reverse a thousand steps of added noise.
McDonald's spent nearly five years building an AI to take drive-thru orders, then switched it off
It sold the system to IBM in 2021 and ran it at more than 100 US restaurants. McDonald's never published how often it got an order right, though reporting guessed about 85 percent.