Given how LLM-based "AI" works, if Claude tends to write regex based code parsers that means that a large majority of (human written) code out there is of limited capability or buggy or just plain wrong.
This also gives a mechanism for how vibe coding could introduce really inexplicable app behavior - regexes that work correctly a lot of the time but not always. Looking forward to it!
This also gives a mechanism for how vibe coding could introduce really inexplicable app behavior - regexes that work correctly a lot of the time but not always. Looking forward to it!