The Death Of Grammar Ronin

One of the things that comes with writing for a public audience in the digital age is the editor without portfolio. This is the person who roams the internet looking for spelling errors, punctuation mistakes, and grammar issues. There are many of these people, as the comment section of every internet post has at least one comment about a typo or alleged improper word choice. They are like the samurai without a master in feudal Japan, except they wield the blue pencil instead of a sword.

Soon, of course, they will be replaced by AI. It will not be long before the browsers simply rewrite your text in the period between when you hit submit and the text commits to the website. The robots will patrol the internet like the grammar ronin of this age but do so with a speed that the grammar ronin cannot match. Imagine a terminator sent back to seventeenth-century Japan to battle Miyamoto Musashi. By the looks of it, the days of the grammar ronin are numbered.

At least it seems that way if you assume there is only one way to construct a sentence or that the rules of grammar are iron laws of grammar. That is often how the grammar ronin look at language and writing. The rules of grammar are not merely guides to facilitate clarity but laws that must be ruthlessly enforced. Even if the grammar rule no longer works for a modern audience, the grammar ronin insist that it must be followed lest chaos be unleashed on humanity.

It turns out that this is where AI disagrees with the grammar ronin. If you compose an essay and feed it into each of the AI, asking for corrections of spelling and grammar, the result will be different from each AI. If you submit the output of one into another, it will rewrite the text for what it claims is clarity and convention. You can create a game of telephone with the AI editors, and when you get to the last one and submit it back to the starting AI, the result is nothing like where you started.

It seems that the robot editors cannot agree on the rules. The reason for that is the AI learns on the mass of text made available to it. Once the robot is seeded, it then continues to build its knowledge based on available information from the internet and what has been fed to it by users. What we call AI is actually a massive probability calculator that quickly returns the most likely answer to the user query, based on the data that has been made available to it.

This is why the results from each AI are slightly different when they are asked to edit the exact same text. There are small differences in its massive data sets, so the probabilities are slightly different. Ask each AI to add simple numbers, and the results are uniformly the same because probability plays no role in the result. Two plus two equals four for all possible values of two. Ask each AI to edit this paragraph, and the range of possible answers is quite broad.

That is because those laws of grammar that the grammar ronin enforce are not laws after all but merely a set of conventions. In fact, what we think of as the rules of grammar are mostly the result of the printing press. Formalizing the language was a natural consequence of the mass production of text. Printers needed to be trained, and therefore it made sense to have a common set of rules. It is how we got the word stereotype, for example.

Of course, the reason we have things like grammar rules, punctuation, fixed definitions, and formal spelling of words is clarity. Many of the punctuation marks we commonly use were relatively late additions to our language. They were created by monks and scribes to make their lives easier. Dictionaries were created to make written communication easier. The iron laws of grammar and spelling are not iron laws after all but things we invented as needed.

This is where AI can be liberating. What the robots can do for the writer is offer many ways to phrase something and then let him select that which fits his style or that he thinks gets his point across the best to the human reader. At the same time, it can also allow the writer to break convention by seeing the conventional ways AI presents the text and then deliberately choosing an unconventional approach. Creative writers can use AI to enhance their creativity.

On the other hand, when a pusillanimous popinjay takes issue with a point a writer is making but is unable to follow the logic that reaches that conclusion, so he attacks the grammar of the writer, the writer can simply point to the terminator and say, "take it up with my editor." Having AI as an editor provides an authoritative defense against this sort of pedantry that is popular with the sophists. In a way, AI can become something like a universal style guide for the digital age.

It is not all rainbows and puppies. The grammar police have drained a lot of the life from the written word, and AI will help them bleed it white. In time, most people will rely on AI to write their text, and that means it will narrow to the point where most writing reads like the user manual for your toaster. This will also make stupid people seem less stupid, which is a great danger to society. This is the problem with politics. It is dominated by loquacious simpletons.

The main loser in the AI revolution will be the grammar ronin. Soon, they will not be able to find text that violates their interpretation of Strunk and White. If they persist, the robots producing the text will simply disconnect them from the internet, leaving them to roam the countryside with a blue pencil in search of bits of paper to edit. The era of the grammar ronin is coming to an end. He will be defeated by the thing that made him possible at the dawn of the internet: technology.