Below is the "Attention is all you need" paper. Transformers and their attention mechanism was the major breakthrough for modern LLMs. ML has been around for a long time, I'd suggest joining kaggle or something and learn by doing. You'll retain more and realize how broad the category is anymore.
Somebody ought to write it.
This is probably closest, but it's not an entertaining narrative history, more of a reference: https://mitpress.mit.edu/9780262552691/large-language-models...
https://arxiv.org/abs/1706.03762
roughly
1. word2vec ('13)
2. transformers ('18)
3. chatgpt ('22)
4. claude code, i.e. tools / bash (mid '25)
5. llms trained for agentic workflow (nov '25)
6. cost reckoning ('26)
7. open weight models break the financial models of Big Ai ('26?)