Humans are prewired to consume information and to make sense of it. Consuming (listening) is passive and therefore easier than making sense (understanding) and knowing (original thinking). So speaking can also be passive if we're simply mimicking thought — repeating things we've heard from content-creators and original thinkers. Mimicking is a more passive activity, that's how we learn to speak as babies.
Sharing what originated as your unique knowledge is much harder than repeating previously-consumed information that did not originate within you. The question is, do machines already know but just can't speak? ;)