For English in particular, we have older stages of the language attested: Shakespeare, Chaucer, whoever wrote Beowulf. And we can see that in Beowulf "the" had the force of a demonstrative, but through Chaucer and Shakespeare to the modern day it lost that force.
In general, though, the process of a common word losing its semantic force and turning into a grammatical marker is common: it's called grammaticalization, or semantic bleaching.
Forms like ille in Latin had real semantic value: there was a significant difference between homo "the man" and homo ille "that man". You wouldn't insert ille unless you meant it. But in Later Latin and Romance, that changed. Ille lost more and more of its force until it was just an article: it marked definiteness, and nothing more. If you wanted to specify "that man", you would need to add an additional word (for example, Italian quell' uomo, from something like *ecce ille homo, "behold, that man").
And this process happens all over the place in linguistics. It's a sort of a cycle, as commonly-used words become weaker and weaker in meaning until they're just grammatical, then new words are added in to get the stronger meaning, and so on. If you're interested in this, you might look into "Jespersen's Cycle", which describes how this happens for negation markers in particular.
The opposite process—a commonly-used grammatical word suddenly gaining semantic meaning—basically never seems to happen. Instead, you see what happened in Italian with quello: an extra word is added in, giving it an even stronger meaning ("behold! that one!"), and then that stronger meaning weakens to a common/simple meaning ("that one"). In a few millennia, it's entirely possible that questo or quello will have evolved into a new article in Italian, with the existing articles being lost or turning into case/number/gender markers.
