LLMs: Loss of Logical Minds
In my earlier post, I talked about how authenticity is an effective tool against LLMs and it got me thinking about how my own writing changed since ChatGPT got mainstreamed three years ago. First, I tend to write less as automation has made it extremely easy to generate every day texts. Second, the act of writing itself became pointless because it felt that every writing exercise became the foundation of for another datapoint for LLM's (to illegally) train on. Third, my overall vocabulary for things has shrunk vastly than the previous years, which intersects with lesser time spent reading and more time spent watching short form videos and interacting with internet speak online, especially on social media.
I want to address the last point specifically, where the way we use language on social media is not the way we'd use words otherwise, optimized for maximum engagement which means a shrinking vocabulary and speaking the zeitgeist which effects the way we think. After we are the words we use and if this part shrinks so is our ability to express complexity and revert to black/white thinking. See any kind of political nuance on a social media platform of your choice and it eventually transforms to right wing skullduggery devolving into name-throwing and simplistic narratives and if anyone tries to counteract them with nuance then they're presented with the sleight of being intellectual.
In my own experience the ability to write is a way to think, and the removal of this skill from the everyday means that I think less and my brain gets used to not putting effort in seeking information. I no longer do much mental math, go hunt sources, read a whole book when all the information I need is served to me as a digestible glop, like a mass market newspaper going for the dopamine jugular. So I have ever more information but lesser ways to truly figure out what I think and where I stand, which also removes me from the democratic decision making loop where I see any information as being absolute (there is no way to edit LLM based information atleast on the user end or report it being false) and go with the flow rather than finding my voice and expressing it.
So I am exposed to an ever more fatty version of information, one that hooks and titillates but that doesn't want me to engage or question. I ask myself, whose purpose does that serve? Not mine definitely, if dear reader you're reading this, it doesn't serve yours either.