Gotcha, I read the copy and hopped into the comments too quickly.

These LLMs are goofy when approached like a vending machine.

They're making assertions in all directions depending on who they're responding to and sources gathered and synthesized for that particular query.

Front-loading context through prompt-stacking makes them intelligent thinking partners.

Otherwise, they seem to ebb and flow with internet sentiment.

Black and Jewish pride must be gen pops' jam these days.