This was supposed to be an article about how LLMs are now ingesting their own output to a certain degree. Which might be true, I mean I generate myself LLM-slop at ark which in turn gets processed by LLMs, and there we have it: my own snake eating itself.
Lately I'm quite conflicted about this, not only because I'm part of the whole Ouroboros dilemma, but this is where my old absurdist views à la Albert Camus keep me sane to a certain degree. In the sense that on one side LLMs are awesome for levelling up my knowledge, not only about coding, but also bits that before were quite opaque to me. All of this whilst by using them I'm feeding them fresh filtered, original meat-bag content.
For example, I remember back in uni, when my OOP design patterns professor kept reminding me that LISP and its dialects are like the real holy grail of programming. My first interaction with a LISP dialect was the AutoCAD's scripting engine, and it looked indeed like an alien language.
Let me avoid digressing too much: LLMs are helping me to finally start grokking the "hard" bits that my brain kept skipping about LISPs: like macros and how they are "homoiconic". Before this, books like Seven Languages in Seven Weeks did the job, but it did not click for macros, just for how their elegant baked-in concurrency/parallelism works.
So yes, LLMs bring back that subtle angst, that absurdist view of the world which you either accept or you let it eat you slowly from the inside. Just to qualify how I use absurdism here i.e. not in the correct sense of the meaningless of life, but in a sub-set: the meaningless of knowledge, yet at the same time thou stills seek it.
Since existence itself has no meaning, we must learn to bear an irresolvable emptiness
Since knowledge itself has no meaning, and LLMs are making it even emptier - we must learn to bear an irresolvable emptiness whilst still consuming it.
Sources