Can AI-enhanced humans discover L[1/4] factoring algorithm?
There are certain rumours in certain corners of the internet that NSA/GCHQ have an L[1/4] factoring algorithm, i.e. they may be able to beat the number field sieve, at least in theory.
Assume for one second this is true, or if not, at least that L[1/4] exists.
Many people say that AI is just a next-word guesser etc, but there is no doubt about its usefulness in presenting relevant material before the eyes of a human.
No-one has publicly discovered L[1/4] factoring despite half a century of intense research. Publicly, humans thus far have failed on their own.
Can AI-assisted research conceivably push the boundaries just a bit beyond the collective genius of human mathematicians, through e.g. time saving in experimentation/scripting, connection of ideas, judicious human postulation with dead ends abandoned and promising avenues explored more quickly than ever before, and find that L[1/4]?