Map the Wild

I was reading an article titled Why Some People Struggle in a Coding Bootcamp and Others Don’t and because I’ve definitely danced the code “dance” before it caught my eye.

So what is this “dance”? It’s when the hot topic, the zeitgeist, the what’s “in” of capitalism starts spinning its consumerist flywheel of courses, events, books, podcasts, and trainings touted by titular talking heads all crawling out of the wood work. It happened with web 1.0, the 2.0, the 3.0, the industrial revolution 4.0, the blockchain, the macarena, don’t tell me you missed the macarena? gasp, clutch pearls, etc

Anyway, after absolving me of my feelings of coding inadequacy (he said I might be too smart to code, even if you read that differently, I’m sticking to it) the article made me think through what machine learning terms would look like, in terms of analogies. Alright, I wanted you to read it yourself, so here’s what he said:

“communication professionals are masters of picking up non-verbal cues about why people are ticked off, why are they excited, what would it take to get them going, and/or address their unspoken needs. So as you can see, those people are masters of the natural communication realm!”

So here’s some AI assisted analogies for what the different terms I keep coming across.

Traditional Language Models: A person relying solely on their memory and knowledge without consulting any external sources.

RAG (Retrieval-Augmented Generation): A person visiting a library to look up information before answering a question.

RLHF (Reinforcement Learning with Human Feedback): A student improving their answers over time by receiving feedback from a teacher.

Fine-Tuning: A student in a classroom, intensely studying specialized books and doing targeted practice exercises, with a teacher guiding them.

Zero-Shot Learning: A person solving a completely new type of puzzle using their general knowledge, without any prior training.

Few-Shot Learning: A person learning to perform a new task after being shown a few examples, like assembling a piece of furniture with a brief tutorial.

Transfer Learning: A person applying skills and knowledge from one field (like problem-solving in math) to another field (like physics), seamlessly transferring their expertise.

Prompt Engineering: A teacher asking carefully worded questions to a student, who then provides more precise and relevant answers, showing effective guidance.