“Unfortunately, human beings assume that LLMs are sentient and understand the questions they're asked, rather than being unthinking statistical models that coug

“Unfortunately, human beings assume that LLMs are sentient and understand the questions they're asked, rather than being unthinking statistical models that cough up the highest probability answer-shaped object generated in response to any prompt, regardless of whether it's a truthful answer or not.” I don’t think this is something that will stay true. Users become more educated on what their systems can and cannot do as well as an intuition of their inner workings over time. Getting the shape of the highest probability answer is a very useful tool that shouldn’t be underestimated, especially if it’s understood that way. As for the Recall related stuff in the rest of the article, I have similar sentiment. Only difference is that I feel like they could really pull off a very useful, similar tool but there are definitely a lot of serious concerns to be solved here first, and rolling it out in its current state is morally irresponsible of those pushing it at Microsoft.