"zero-shot-cot-grok": how bizarre modern LLM usage looks from a 2019 data science view

gleech · x · 2026-09-29

The author muses on how bizarre today's AI jargon and practices look from a Data Science of 2019 perspective — the term "zero-shot-cot-grok" alone (zero-shot + chain-of-thought + Grok) would have been incomprehensible just a few years ago.

In just a few years, models doing complex reasoning with zero-shot prompting has become the default — and the speed of that paradigm shift is itself remarkable.

Original post →

More from Fun

Fun channel →