pointing out to a coworker that there's no meaningful technical counter for LLM semantic drift because it's inherent in the training data - people use it all the time when they're trying to disingenuo

pointing out to a coworker that there's no meaningful technical counter for LLM semantic drift because it's inherent in the training data - people use it all the time when they're trying to disingenuously win arguments, for example.