The Revenge of the High School English Teacher

Do you remember high school English where your teacher seemed to arbitrarily like or dislike your essays? Remember comments such as “word choice?” or “sweeping generalization!” or or even more cringe-worthy, “off-topic!” You of course as an all-knowing high schooler thought your words were perfect and your teacher clueless and by golly, you’ll never take another English class again if you can help it.

Your teacher patiently, resignedly (he’s been at this for at least 20 years) explains to the class that both what you write with words and how you express it are immensely important. Little words, such as “to” or “from” — also known as prepositions in the classical world— make all the difference in meaning. He gives you examples of great works of enduring writers such as Nathaniel Hawthorne and Virginia Woolf. He asks you to study exemplary speeches written by Patrick Henry and Martin Luther King. Your high school self recognizes the power of those words but you dismiss them as one-offs or archaic. Meanwhile you really don’t want your writing to be judged by your English teacher or anyone else for that matter.

So, you go to university and major in a technical subject such as computer science where you’ll never again have to put up with a capricious teacher who has the immense gall to judge your inner most thoughts translated into words.

Oh, you technical, sweet innocents! All of your years spent calculating linear algebra equations and writing obtuse C++ code has brought us to this advanced technological point where you have finally managed through collective efforts to achieve Artificial Intelligence — not in the form that Turing expected — but in something even more impressive: machines that simulate an understanding of our natural language aka Large Language Models.

We’ve all studied how this development came to be through a confluence of factors such as GPUs developed for gamers, steady advances in machine learning algorithms, and new approaches to training. Yes, it has taken brilliant technical and mathematical brains to reach this point.

The irony? Now you don’t need to know how to program (goodbye C++ and python) to “program” our large language models to do work for you. All you need is a solid grasp of English.

P.S. Words matter now, more than ever before.