I don't believe OP's thesis is properly backed by the rest of his tweet, which seems to boil down to "LLM's can't properly cite links".

If LLM's performing poorly on an arbitrary small-scoped test case makes you bearish on the whole field, I don't think that falls on the LLM's.