Anecdotal evidence suggests that LLMs are quite content to use atomic variables for custom multithreaded signaling and synchronization logic even when safer alternatives like mutexes or messaging channels are available.
LLMs aren't creative, they're statistical, this is axiomatic. That means the use of atomic variables instead of mutexes is endemic to human written code out there.
I find that odd. Does your average programmer just not care about thread safety?
LLMs aren't creative, they're statistical, this is axiomatic. That means the use of atomic variables instead of mutexes is endemic to human written code out there.
I find that odd. Does your average programmer just not care about thread safety?