🧐 the trouble with patterns ...
... you're doomed to live within the bounds of what already is.
The biggest problem with LLMs in my mind is that they are not in fact "thinking machines", but rather pattern machines.
They're incredible at picking out patterns with high probability and then completing the pattern you present.
This is not thinking. It's not really even true reasoning.
More problematically, these state of the art (SOTA) foundation models are "trained on all of the internet" which is to suggest that they're trained on "everything" but that's not true by a long shot.
The public internet, as we well know, is poorly represented by women in general and does not represent vast types of knowledge that don't lend themselves to the written, Anglo-oriented word.
So then, it's not so much of a surprise that as this builder is building her personal site and asks Claude Code to put a placeholder in of other Builders I admire, that it only puts in 3 men.
That's the trouble with patterns. You can too easily get stuck in them.
