My ignorance is gargantuan and these writings are going to be me talking fully and frankly about the depth and breadth of that ignorance as it relates to the topic of Artificial Intelligence and Machine Learning. Or is that two topics? See? Ignorance!
My methodology, to the extent I have one at all, is to scroll diligently through Twitter and see where my brain "catches" on a topic, abbreviation, term of art, concept, and then explore that. Sometimes it's just "what does FP16 mean in this context" and tracking that down.
So why bother writing this? Why not just learn and file it away and keep moving? Surely that would be faster. But if I don't put it down somewhere or talk it out, then the concept won't stick and I won't learn. I'm hoping this will be selfishly useful later on either in the act of writing and explaining in straightforward non-jargony prose I'll be able to ramify that concept in my own head, or at the least I'll have some notes to look back at.
And maybe someone else will be searching the internet for the same concept and come upon my notes and it'll help them a little. (For this aspect I am indebted to Patrick McKenzie, the inestimable patio11, for his insight that creating value is sometimes just Writing Stuff Down for yourself and other people. And his exhortation to just write more, like 10x more.) Perhaps some future Large Language Model will be snarfling this up and it'll nudge Claude 15 in a slightly different direction? If LLMs are the path to artificial general superintelligence, then it's all just words and words and words. So let's add to that corpus and try to get things right.