As a technology journalist and data analyst, large language models (LLMs) pose a conundrum. On one hand, they are based on the taking of people's work to create knowledge machines that are not for the public benefit, but for to enrich a few. On the other hand, when used for specific tasks and with full knowledge of their deficiencies, LLMs can save time and improve quality.
That said, here is how I use the current crop of large language models:
- I do not use LLMs for writing or editing my work.
- I do not use LLMs for art work (aside from the site logo and the anime-inspired image on this page).
- I occasionally use LLMs to help with research.
- I always check anything provided by an LLM, by finding reputable sources to verify provided facts.
- I frequently use LLMs as a pair programmer and assistant during my data science work and coding.
- I frequently experiment with new ways to use LLMs.
(More to come.)