100 pages vs 11 words.

How difficult is it to find a solution for something like this: “Mary baked some breads. For each bread, she boiled the same (as the bread) number of eggs. For each egg, she made the same (as the bread) number of pickles. If the total number of food items on the table is 39, how many breads did she bake?”

The answer is not immediately obvious. Indeed, word problems in these forms are not only long-winded but also prone to misinterpretation. I may have wanted to ask you something but you may have understood (and answered) something else entirely. Moreover, I may have misspecified what I originally wanted to ask. All in all, this seems like a rather unfortunate way of asking questions of such nature. What if I change it to something like this instead:

Solve for x if x^3 + x^2 + x = 39. (Writing on mobile, so sorry for this not being properly formatted). Much better. Not only is this formulation succint, but it’s also much more unambiguous. And, in fact, it’s understandable even if you are not a native English speaker.

This is what’s known as an Abstraction. What I’m asking you remains unchanged but now instead of using long sparsely-informative symbols, I’ve compressed the entirety of it to an utterly irreducible form, negating any chances of miscommunication of ideas between you and me. Let me come to the point, in case you’re still confused about where this is going. Language (as we use them right now) are not an entirely ideal method for communication. They are ambiguous, superfluous, non-universal, sparse, and sometimes ridiculous (apparently, all fax no printers, is a semantically correct sentence these days). Certainly, for years, they have been changing steadily, depending upon culture, climate, geography, history etc. but still have somehow managed to remain inefficient. But now that will change drastically. And yes, you guessed it, because of Language Models.

People generally lament the fact that Language Models exist. And I empathize with them. Command over language used to be a strong indicator of intelligence, much like how a well-decorated cake is a strong indicator of its taste. But not anymore. Language Models have devastatingly isolated intelligence and language to the extent that the cake can (and generally will) taste horrible even if it looks expensive. However, I want to look at the bright side of the things. In a world where all cakes look fabulous… taste will have to be all that matters, no? Which is to say, when there’s no reward in writing beautiful prose (except for artistic endeavors), won’t idea be all that matters?

And as we all know, ideas are generally much succinct than the means used to convey them. Which gives me hope that we’ll inadvertently gravitate towards better abstractions than the traditionally trite prose. I don’t know how language will evolve to be honest but I am pretty sure that such evolution would be beautiful and not mechanistic. Because, after all, we have existing examples amidst us. Why, what can still not be constrained about life, loss, love, and hope in more than 100 something pages can beautifully be summarized in mere eleven words:

Would you know my name, if I saw you in heaven.